Temporal integration definition

Temporal integration definition DEFAULT

Open Access


  • Scott L. Fairhall ,
  • Angela Albi,
  • David Melcher
  • Scott L. Fairhall, 
  • Angela Albi, 
  • David Melcher



Figure 1
Figure 2
Figure 3


There is increasing evidence that the brain possesses mechanisms to integrate incoming sensory information as it unfolds over time-periods of 2–3 seconds. The ubiquity of this mechanism across modalities, tasks, perception and production has led to the proposal that it may underlie our experience of the subjective present. A critical test of this claim is that this phenomenon should be apparent in naturalistic visual experiences. We tested this using movie-clips as a surrogate for our day-to-day experience, temporally scrambling them to require (re-) integration within and beyond the hypothesized 2–3 second interval. Two independent experiments demonstrate a step-wise increase in the difficulty to follow stimuli at the hypothesized 2–3 second scrambling condition. Moreover, only this difference could not be accounted for by low-level visual properties. This provides the first evidence that this 2–3 second integration window extends to complex, naturalistic visual sequences more consistent with our experience of the subjective present.

Citation: Fairhall SL, Albi A, Melcher D (2014) Temporal Integration Windows for Naturalistic Visual Sequences. PLoS ONE 9(7): e102248. https://doi.org/10.1371/journal.pone.0102248

Editor: Virginie van Wassenhove, CEA.DSV.I2BM.NeuroSpin, France

Received: April 11, 2014; Accepted: June 17, 2014; Published: July 10, 2014

Copyright: © 2014 Fairhall et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

Data Availability: The authors confirm that all data underlying the findings are fully available without restriction. All relevant data are within the paper and its Supporting Information files.

Funding: The research leading to these results has received funding from the European Research Council under the European Union’s Seventh Framework Programme (FP/2007–2013)/ERC Grant Agreement n. 313658. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Competing interests: The authors have declared that no competing interests exist.


The brain integrates incoming sensory information not only over space and sensory modality but also over time. Due to the diverse nature of this information, temporal integration mechanisms may vary across different timescales. The integration of environmental events into a unitary percept may occur within a few hundreds of milliseconds [1]. However, the integration of more complex information that unfolds over time may utilize neural mechanisms that operate across longer time scales. While varied mechanisms may operate over periods as long as minutes [2], [3], evidence has accumulated to suggest that temporal integration windows (TIWs) of 2–3 seconds may be a fundamental component of human cognition [4], [5].

One of the main tasks of the perceptual system is to parse the continuous input from the senses into meaningful objects and events. This goal necessarily involves integrating information over time. In the case of a real-world objects, for example, the influx of relevant information coming from the different senses at different sensory delays, such as vision, audition and touch, are combined over time in order to define the object as a unique spatiotemporal entity. Individuating and recognizing objects is thought to take on the order of 150 to 200 milliseconds [1], [6]. The fact that information is combined over time can be seen in judgments of simultaneity [7], [8] as well as in visual masking studies in which a target and mask are perceptually combined even though they are discrete events. A useful metaphor for describing these periods in which sensory input is combined is a “temporal integration window” [4], [5]. If two stimuli fall within the same temporal window then they are integrated into a coherent percept, while if the two stimuli fall in different windows then they are perceived as separate objects/events [1], [9].

Perception of complex phenomena such as motion [10], apparent motion [11], [12], biological motion [13] and events [2], [3], [14] requires temporal integration over longer periods of time because these entities are, by definition, extended in time. In the case of apparent motion, for example, if two brief stimuli are separated by less than a few hundred milliseconds, observers tend to perceive smooth and continuous motion in between the two discrete stimuli [11], [12]. In particular, evidence has accumulated to suggest that temporal integration windows (TIWs) of 2–3 seconds may be a fundamental component of human cognition [4], [5]. This 2–3 second integration window appears to be a ubiquitous feature of human cognition rather than being specific to particular cognitive processes or perceptual/motoric contexts. Both auditory and visual temporal intervals can be reproduced with high fidelity and with little across-trial variability up until 2–3 seconds, before the capacity to accurately represent these intervals breaks down [15], [16]. Likewise, the capacity to produce precise anticipatory motor actions synchronized with predictable auditory cues fails as inter-stimulus intervals exceed 3 seconds [17]. This phenomenon reflects more than a simple timing mechanism. The accumulation of evidence that allows the detection of motion coherence embedded within noise also asymptotes around 2–3 seconds as the integration mechanism reaches capacity [10], [18]. Moreover, the 2–3 second TIW extends not only to perception but also language and motor production. Speech utterances have been reported with a duration clustered around 2.5 seconds [19] and cross-cultural ethological studies have documented that the natural performance of motor actions is segmented within a 2–3 second window [20].

The prevalence of the 2–3 TIW across modalities, tasks, perception and production has led to the suggestion that it may reflect a general organizing principle of human cognition - better defined as the ‘subjective present’ i.e. the phenomenal impression of ‘nowness’ [5], [21]. This is an intriguing possibility but an important test of any such claim is whether this TIW can be observed in the processing of stimuli that more closely match our subjective experience. Normal perceptual experience involves a rich and tumultuous barrage of information while processes such as the reproduction of a tone or the accumulation of coherence information do not. In order to more closely approximate the complexity of real life, movie clips may act as a useful, if not perfect, proxy for our day-to-day subjective experience. In particular, movies typically contain multiple objects, motion-paths and events as well as shifts of attention (and gaze) as stimuli become more or less salient. If the function of the 2–3 second TIW is to integrate complex sequences of events into a coherent conscious stream then movie-clips present an important test of whether this integration window is an aspect of our subjective experience.

The goal of the present study is to address whether the 2–3 second temporal integration window extends to complex stimuli more consistent with our subjective experience. According to the logic of a temporal window, the brain should be able to combine information within the limits of a single TIW, even when the order of that information is scrambled, but have difficulty when information is scrambled over longer timescales. A similar approach has been used to identify differences between native and non-native speakers in integration capacity for phonetic sounds in spoken language [22]. We used movie clips and temporally shuffled the sequence of events over a range of different scales, from a few hundred milliseconds to several seconds. We hypothesize that if the 2–3 second TIW is indeed critical for understanding events in the subjective present, then there should be a dramatic increase in the subjective impression of the difficulty of following the movie as the duration of the window of temporal scrambling increases beyond the 2–3 second time period.



There were 15 participants in Experiment 1 (mean age: 23.4, 12 female) and 28 separate participants in Experiment 2 (age: 24.1, 18 female). The numbers were predetermined to allow complete counterbalancing of the videos in each condition. All gave written informed consent and received a small monetary compensation for participating. This study was approved by the Ethics Committee of the University of Trento in accordance with the provisions of the World Medical Association Declaration of Helsinki.


On each trial, stimuli consisted of 12.8 second videos (90 in experiment 1, 98 in experiment 2). Video segments were selected from a pool of 7 relatively obscure international movies and the audio component was removed. First, 424 random clips were selected. One experimenter (A.A.) rated each video on a seven-point scale for the presence of a simple narrative. Of these, the best 100 were selected. Ninety were used in Experiment 1, 98 in Experiment 2).

Videos were presented at a frame rate of 25 Hz and a resolution of 360 by 272 pixels (subtending approximately 28° horizontally and 21° vertically). In order to better control for differences in low-level visual change across conditions, we introduced a visual transient every 5 frames. This manipulation was necessary because temporal scrambling in itself produces transients. By introducing the transient uniformly every 200 msec it is possible to balance the overall occurrence of transients across shorter and longer temporal scrambling intervals. Thus, across all conditions, videos were divided into base-units of 5 frames (i.e. 200 msec) where the 4th frame was faded (RGB values halved) and the fifth frame was replaced by a blank frame.

Stimuli were shuffled within different time windows in order to see whether there was a discontinuity in perception when TIWs exceed around 2 seconds. To manipulate temporal integration demands, a segment of a fixed duration was taken and segments of the video were shuffled within this time window. For instance, for a 1600 msec TIW, sections of the video within the first 1600 msec were shuffled across time. Then the process was repeated across the next 1600 msec until the end of the video. Temporal shuffling was random with the exception that no segment followed its original predecessor.

To manipulate the overall level of shuffling independently of the TIW duration, we introduced the concept of a ‘shuffle-chunk’. Videos were shuffled within TIWs to different extents. For instance, in the 1600 msec TIW example, videos might be shuffled in either 200 or 400 msec chunks – thus the shuffling could be twice as frequent in the 200 msec condition (8 segments) than the 400 msec condition (4 segments) while preserving the overall duration of the TIW. This manipulation was important in order to vary the overall amount of shuffling and the temporal integration demands.


Experiments were run on a PC computer and controlled by Matlab (The MathWorks, Inc., Natick, Massachusetts, United States) using Psychtoolbox [23], [24]. Each video clip was presented only once to each participant (Experiment 1: 90 trials; Experiment 2: 98 trials) in one of the TIW/shuffle-chunk combinations. Videos were counterbalanced across subjects such that each video was seen an equal number of times at each level of the TIW/shuffle-chunk factorial design. The number of trials per participant within each cell of the factorial design was 6 in Experiment 1 and 7 in Experiment 2. We collected a subjective rating [25], [26], [27] of the subjects’ impression of how effortful the video was to watch. Specifically, participants rated the difficulty to follow (DtF) of each movie clip on a nine-point scale, where 1 indicated easy to follow and 9 indicated very difficult to follow.

To encourage vigilance, on 25% of trials the subjects also reported the basic narrative of the video using an open-ended keyboard response. These open-ended responses were not analyzed.

Normalization of Ratings Across Participants and Videos

To standardize individual differences in usage of the rating scale, all responses were normalized such that each subject had a mean of 0 and a standard deviation of 1 across all responses made by that subject. In addition, to normalize differences in the innate difficulty to follow of each video segment, ratings of the same video across subjects were normalized such that each video had a mean rating of 0 and a standard deviation of 1.

Exclusion of Potential Confounds

In order to allow for more naturalistic stimuli, we used movie clips rather than highly controlled but artificial stimuli. The videos were selected to be representative of the types of video clips we generally view and contained a wide range of cuts (between 0 and 11) with a mean shot-duration of 2.78 seconds and a broad distribution (standard deviation: 4.96 seconds). Inevitably, confounds may arise that interact with condition of interest. For example, visual change was likely to increase with the length of TIWs, as increasingly disparate sections of a video-segment are placed next to one another. We calculated the variation of potential confounds with our experimental manipulations for use in subsequent analyses.

To determine visual change, videos were down-sampled to a 5 vertical by 7 horizontal grid. This matrix was then vectorised and, for N-1 frames, each frame was correlated with its succeeding frame (skipping the blank frame occurring every 5th frame). The result gave an acute measure of visual change (1-r). A log linear relationship was seen between TIW duration and degree of visual change (Exp 1: R2 = .21; Exp 2: R2 = .41). An approximately linear relationship was seen between cluster-chunk duration and visual change (Exp 1: R2 = –.56; Exp 2: R2 = –.60).

The interaction between experimenter cuts and ‘natural’ cuts (termed cross-cuts) in the video was determined using the visual change process above. An inter-frame threshold of r<.6 was capable of detecting cuts of both a natural and experimental origin while being insensitive to camera pans. This process revealed that a log linear relationship was again present between TIW and cross-cuts (Exp 1: R2 = .35; Exp 2: R2 = .65). A linear effect was seen between shuffle-chunk and cross-cuts ((Exp 1: R2 = .36; Exp 2: R2 = .17). Changes in mean luminance were not correlated with either independent variable (p-values>.2).

Effects of visual change and cross-cuts were controlled using mean adjustment [28]. First, the covariance at the group level between visual change and DtF at each level of the factorial design was determined. Mean visual change for each condition (e.g. TIW: 1600, shuffle-chunk: 400 msec) was determined across video. Covariance was accounted for by regressing each condition against the mean DtF for that condition averaged over participants. The variation accounted for by this potential confound was removed from the data [i.e. DtFadj = DtF-b(VCij–mean(VC)); where b is the regression coefficient between DtF and visual change]. This process was repeated on the adjusted data now using the variable cross-cuts (accounting for potential colinearity between visual change and cross-cuts covariates).

Data Availability

Individual subject and video data are available as supporting information (File S1).


Experiment 1

As hypothesized, the difficulty to follow (DtF) ratings were influenced by the time window of shuffling. A two-way repeated measures ANOVA revealed a significant main effect of the factors TIW (F(4,56) = 14.5,  = .23, p<.001) and shuffle-chunk (F(2,28) = 14.0,  = .12, p<.001) but no interaction between them (F<1). The pronounced increase between 1600 and 3200 msec in DtF can be seen in Figure 1 (T(14) = 4.2, p<.001, Cohen’s d = 1.07, one-tailed). In terms of our 9-point scale, effects were modest. The grand averages for the five TIWs depicted in figure 2 extended over a ∼1.5 point range of the original 9-point scale. However, in terms of relative changes across the TIW range, the increase in DtF between 1600 and 3200 was pronounced - approximately one and a half times (146%) the next greatest increase (that occurring between 6400 and 12800 msec; see figure 2). This effect became more evident once DtF ratings were mean-adjusted using the covariates visual change and cross-cuts. Following adjustment, only the hypothesized difference between 1600 and 3200 msec remained significant (T(14) = 3.37, Cohen’s d = 0.87, p = .004, one-tailed–see dotted line).


Figure 1. Example of a video re-sequencing for a 1600 msec TIW, 400 msec shuffle-chunk condition.

Across all conditions, the base-unit was 5 frames (200 msec). For 400 msec shuffle chunk conditions, two sequential base-units would be combined. In this example, a 1600 msec TIW would thus consist of 4 re-ordered shuffle-chunks. Finally, for 1600 msec TIWs, this process would be repeated 8 times to produce the entire 12.8 second video.



Figure 2. Difficulty to Follow (DtF) as a function of Temporal Integration Window (left) and Shuffle-Chunk (right).

Data are presented both before and after mean-adjustment for low-level visual features (visual change and cross-cuts). Note the pronounced increase in difficulty to follow rating between 1600 and 3200 msec that persists after adjustment for the low-level visual properties of the movie clips. Also note that the effect of shuffle-chunk (the degree of temporal shuffling within a TIW) can be alternatively accounted for by low-level visual features.


The effect of shuffle-chunk within TIWs is presented in the right panel of Figure 1. It is interesting to note that the approximately linear trend between shuffle-chunks and DtF was fully accounted for by visual change (dotted line). Thus, in contrast to the stepwise increase in DtF between 1600–3200, both the degree of shuffling within windows (shuffle-chunks) and changes in DtF at longer or shorter intervals could be alternatively explained by low-level visual properties.

The first data point in figure 2 contains an unscrambled video condition (800 msec TIW and 800 msec shuffle-chunk). It is plausible that the unscrambled video acts as an outlier distorting the data. However, this appears not to be the case. At the 800 msec shuffle-chunk, the difference between unscrambled and the next highest TIW (1600 msec) is significant (t(14) = 1.96, Cohen’s d = 0.53, p = .036, one tailed) but is less than the difference over the critical 1600 to 3200 msec TIW increase (t(14) = 2.11, Cohen’s d = 0.60, p = .027, one tailed).

Experiment 2

To provide an internal replication and to refine our estimate of the TIW, a follow up study was conducted using TIWs of 1200 2000 2800 3600 and 4400 msec and 200 and 400 msec shuffle-chunks. Additionally, a number of trials with unscrambled and fully scrambled (12800 msec) TIWs were included in order to peg responses over a similar range to Experiment 1 but were not included in the analysis.

Results (see figure 3) were consistent with Experiment 1. Both TIW (F(4,108) = 4.9,  = .07, p<.001) and short-shuffle F(1,27) = 20.9,  = .08, p<.001) had a significant influence on DtF and there was no interaction between these factors (F<1). The hypothesized stepwise increase in DtF was again apparent between 2000 and 2800 msec (T(27) = 2.4, Cohen’s d = 0.49, p<.01, one-tailed). This increase was more than twice (220%) the next greatest increase, which occurred between 1200 and 2000 msec TIWs. This increase remained significant after mean-adjustment for the covariates visual change and shuffle-chunk (t(27) = 1.99, Cohen’s d = 0.38, p = .028, one-tailed). The influence of shuffle-chunk did not survive adjustment for visual change.


Figure 3. Difficulty to Follow (DtF) as a function of Temporal Integration Window for Experiment 2.

Data are presented both before and after mean-adjustment for low-level visual features. As in Experiment 1, the sharp increase in DtF between 2000 and 2800 msec persists after mean-adjustment for low-level visual features of the video clips.


Item analysis

The preceding statistical analyses indicate generalizability to the population but would our effects extend to different sets of videos? To address this we reran the main analysis now considering video clips rather than participants as the random factor. The main effect of TI-window was significant (Exp 1: F(4,356) = 14.4,  = .14, p<.001; Exp 2: F(4,388) = 8.3,  = .08, p<.001), Furthermore, as in the preceding analyses the only significant increases in DtF were between 1600 and 3200 msec (Exp 1: t(89) = 2.8, Cohen’s d = 0.29, p<.005) and between 2000 and 2800 msec (Exp 2: t(97) = 3.2, Cohen’s d = 0.25, p<.001). This indicates generalizability to videos sampled in the same manner although it is an open question whether 2–3 second TIW effect would extend to different forms of video (e.g. highly familiar clips, Hollywood trailers).

We additionally considered whether the DtF of the unscrambled version of the video influenced the 2–3 second TIW effect. To do this, we determined the fit of each video to a function modeling a single stepwise increase at the hypothesized TIW. Then we determined the correlation between model-fit and the DtF rating of the unscrambled video. In neither experiment did the TIW effect vary as a function of the DtF of the original clip (Exp 1: R2 = .019; Exp 2: R2 = .034; p-values>.05).


In the present study we investigated whether a 2–3 second TIW operated over the processing of naturalistic visual sequences. In two experiments, we presented 12.8 second video clips manipulated to vary temporal integration demands within and across this proposed window boundary. In both studies, rather that a simple linear increase in the subjective difficulty to follow with temporal scrambling, a dramatic increase in DtF was evident between 2 and 3 seconds. This provides evidence that there is a natural integration window of around 2 seconds that operates even in a very different context to those previously studied, extending to complex multidimensional visual streams more consistent with the subjective experience.

The pattern of responses in both experiments was consistent with a temporal integration mechanism that easily re-integrates sequences within 2 seconds, is progressively strained as the integration window is extended from 2000 to 2800 msec, and then reaches capacity after 2800 msec. This pattern of responses is in accord with previous studies of unidimensional temporal integration. For example, the anticipatory response to predictable auditory cues is relatively flawless until inter-stimulus intervals of about 1200 msec, then progressively breaks down from 1800–3600 msec, after which the anticipatory mechanism fails [17]. Similarly, the capacity to integrate two briefly presented motion coherence fields shows an abrupt reduction in efficiency when inter-stimulus intervals lengthen beyond around 2 seconds [18]. The consistency between these results and the pattern of responses seen in the present study suggests that a similar mechanism (or a similar constraint) underlies the temporal integration of simple unidimensional and complex multidimensional stimuli.

Of note, in neither study was there an interaction between the amount of within-window shuffling (shuffle-chunks) and the effect of TIW duration. For instance, in Experiment 1 there is a linear increase in DtF with within-window shuffling that is independent of the TIW duration. A similar monotonic increase is observed in the across-subject variability of eye movements with scrambled videos [29]. This indicates that two separate mechanisms account for the effect of temporal shuffling on DtF ratings. The first may reflect the overall alteration in the incoming visual sequence and correlate with oculomotor variability while the second independent mechanism instead appears to reflect the time period over which the information must be re-integrated. Thus the 2–3 second integration effect can be seen to be independent of the overall amount of temporal scrambling.

Critically, the 2–3 second temporal integration effect could not be explained by low-level aspects of the videos (visual change and cross-cuts). It is not possible to determine with complete certainty whether the true pattern of results reflects the adjusted (dotted lines in figures 2 & 3) or unadjusted data (gray lines in figures 2 & 3). Low-level visual properties may be only incidentally correlated with our temporal shuffling procedure and irrelevant to the integration process. On the other hand, there may be two complimentary processes at play – one captured by visual change and the other by temporal integration processes. The influence of the short-shuffle chunks discussed in the previous paragraph (also accountable by low-level video properties) suggests the latter case might be true. Specifically, our results suggest that the effect of low-level change on DtF operates independently of the temporal integration process. In either case, both the adjusted and unadjusted data indicate that different processes are occurring before and after the 3-second integration period.

The current findings provide further evidence for a key integration window of around 2–3 seconds [5]. However, studies of working memory span [30], as well as neuroimaging studies of the integration of narrative elements over different time scales [2], [3] suggest that longer time scales may also play an important role in understanding the plot of films. We think it is likely that such long time scales would involve different mechanisms and neural substrates from the 2 to 3 second TIW studied here.

It is interesting to note that although the duration of individual shots in Hollywood films varies greatly, it is rare to find shots less than around 2 seconds [31] (see www.cinemetrics.lv). Movie trailers and action sequences tend to have relatively short shot durations (a larger number of cuts per minute) of around 2–3 seconds (the average shot length in a Michael Bay film is 3.0 seconds: www.cinemetrics.lv), while tracking shots of several minutes can also be found, for example, in the work of director Alfonso Cuarón. Given that people typically move their eyes several times per second, even the shortest shots are usually an order of magnitude longer than human fixation durations while reading. Nonetheless, movie shots have often been compared to fixations by film theorists and directors [32]. This raises the question of why film is so “inefficient” compared to a human fixation and why there are not several cuts per second in typical movies. One possibility, consistent with the current results, is that event information is accumulated over a period of a few seconds, making clip durations of 2–3 seconds an ideal compromise between efficiency (showing as many different shots as possible in a short period of time) and ease of viewing.

In order to perceive coherent objects and events, the brain integrates incoming sensory information over time, over space and across sensory modalities. It is known that temporal integration mechanisms operate across multiple timescales [1]. Integration windows of around 100–150 ms are found in various paradigms, including backward masking and motion integration [4]. Similarly, a number of other studies have reported integration windows of around 300 ms for phenomena such as apparent motion, the attentional blink and inhibition of return [33]–[35]. However, one of the most apparent, yet mysterious features of the stream of consciousness is that there is an integrated subjective present, which has been estimated to extend for around 2 to 3 seconds [5], [21]. Most studies of conscious awareness have focused on much shorter time windows involved in tasks such as detection of a single stimulus. In contrast, the subjective present seems to involve an aspect of consciousness that is extended in time. Here, we examined the role of this time window in our understanding of a complex, multidimensional stimulus that is more consistent with our subjective experiences of the world. Overall, these results suggest that a function of this 2–3 second window may be to provide a stable and coherent representation of events in a complex, ever-changing world.


We would like to thank Timo Stein for his insightful comments on an earlier version of this manuscript.

Author Contributions

Conceived and designed the experiments: SLF DM AA. Performed the experiments: SLF AA. Analyzed the data: SLF. Contributed to the writing of the manuscript: SLF DM.


  1. 1. VanRullen R, Koch C (2003) Is perception discrete or continuous? Trends in Cognitive Sciences 7: 207–213.
  2. 2. Hasson U, Yang E, Vallines I, Heeger DJ, Rubin N (2008) A hierarchy of temporal receptive windows in human cortex. J Neurosci 28: 2539–2550.
  3. 3. Lerner Y, Honey CJ, Silbert LJ, Hasson U (2011) Topographic Mapping of a Hierarchy of Temporal Receptive Windows Using a Narrated Story. Journal of Neuroscience 31: 2906–2915.
  4. 4. Gibbon J, Malapani C, Dale CL, Gallistel CR (1997) Toward a neurobiology of temporal cognition: Advances and challenges. Current Opinion in Neurobiology 7: 170–184.
  5. 5. Poppel E (1997) A hierarchical model of temporal perception. Trends in Cognitive Sciences 1: 56–61.
  6. 6. Wutz A, Melcher D (2013) Temporal buffering and visual capacity: the time course of object formation underlies capacity limits in visual cognition. Atten Percept Psychophys 75: 921–933.
  7. 7. Mach E (1865) On the effect of the spatial distribution of the light stimulus. 1965), Mach bands: 253–271.
  8. 8. Occelli V, Spence C, Zampini M (2011) Audiotactile interactions in temporal perception. Psychon Bull Rev 18: 429–454.
  9. 9. Wutz A, Weisz N, Braun C, Melcher D (2014) Temporal windows in visual processing: “prestimulus brain state” and “poststimulus phase reset” segregate visual transients on different temporal scales. J Neurosci 34: 1554–1565.
  10. 10. Burr DC, Santoro L (2001) Temporal integration of optic flow, measured by contrast and coherence thresholds. Vision Research 41: 1891–1899.
  11. 11. Braddick OJ (1980) Low-level and high-level processes in apparent motion. Philos Trans R Soc Lond B Biol Sci 290: 137–151.
  12. 12. Wertheimer M (1912) Experimentelle studien über das sehen von bewegung: Barth Leipzig.
  13. 13. Neri P, Morrone MC, Burr DC (1998) Seeing biological motion. Nature 395: 894–896.
  14. 14. Zacks JM, Magliano JP (2011) Film, narrative, and cognitive neuroscience. Art and the senses: 435–454.
  15. 15. Elbert T, Ulrich R, Rockstroh B, Lutzenberger W (1991) The Processing of Temporal Intervals Reflected by Cnv-Like Brain Potentials. Psychophysiology 28: 648–655.
  16. 16. Kagerer FA, Wittmann M, Szelag E, Von Steinbuchel N (2002) Cortical involvement in temporal reproduction: evidence for differential roles of the hemispheres. Neuropsychologia 40: 357–366.
  17. 17. Mates J, Muller U, Radil T, Poppel E (1994) Temporal Integration in Sensorimotor Synchronization. Journal of Cognitive Neuroscience 6: 332–340.
  18. 18. Melcher D, Morrone MC (2003) Spatiotopic temporal integration of visual motion across saccadic eye movements. Nature Neuroscience 6: 877–881.
  19. 19. Vollrath M, Kazenwadel J, Kruger HP (1992) A Universal Constant in Temporal Segmentation of Human Speech - a Reply to Schleidt and Feldhutter (1989). Naturwissenschaften 79: 479–480.
  20. 20. Schleidt M, Eibel-Eibelsfeldt I, Poppel E (1987) A Universal Constant in Temporal Segmentation of Human Short-Term Behavior. Naturwissenschaften 74: 289–290.
  21. 21. Poppel E (2009) Pre-semantically defined temporal windows for cognitive processing. Philosophical Transactions of the Royal Society B-Biological Sciences 364: 1887–1896.
  22. 22. Kiss M, Cristescu T, Fink M, Wittmann M (2008) Auditory language comprehension of temporally reversed speech signals in native and non-native speakers. Acta Neurobiol Exp (Wars) 68: 204–213.
  23. 23. Brainard DH (1997) The psychophysics toolbox. Spatial Vision 10: 433–436.
  24. 24. Pelli DG (1997) The VideoToolbox software for visual psychophysics: transforming numbers into movies. Spat Vis 10: 437–442.
  25. 25. MacDorman KF (2006). Subjective ratings of robot video clips for human likeness, familiarity, and eeriness: An exploration of the uncanny valley. In ICCS/CogSci-2006 long symposium: Toward social mechanisms of android science (26–29).
  26. 26. Kortum P, Sullivan M (2010) The effect of content desirability on subjective video quality ratings. Hum Factors 52: 105–118.
  27. 27. Benard MR, Baskent D (2013) Perceptual learning of interrupted speech. PLoS One 8: e58149.
  28. 28. Howell D (2012) Statistical methods for psychology: Cengage Learning.
  29. 29. Wang HX, Freeman J, Merriam EP, Hasson U, Heeger DJ (2012) Temporal eye movement strategies during naturalistic viewing. J Vis 12: 16.
  30. 30. Vandenbroucke ARE, Sligte IG, Lamme VAF (2011) Manipulations of attention dissociate fragile visual short-term memory from visual working memory. Neuropsychologia 49: 1559–1568.
  31. 31. Salt B (1983) Film style and technology: History and analysis: Starword London.
  32. 32. Messaris P (1994) Visual “literacy”: Image, mind, and reality: Westview Press.
  33. 33. Kolers PA (1972) Aspects of motion perception: Pergamon Press Oxford.
  34. 34. Posner MI, Cohen Y (1984) Components of Visual Orienting. Attention and Performance 10: 531–556.
  35. 35. Raymond JE, Shapiro KL, Arnell KM (1992) Temporary Suppression of Visual Processing in an Rsvp Task-an Attentional Blink. Journal of Experimental Psychology-Human Perception and Performance 18: 849–860.
Check for updates via CrossMark

Subject Areas


For more information about PLOS Subject Areas, click here.

We want your feedback.Do these Subject Areas make sense for this article? Click the target next to the incorrect Subject Area and let us know. Thanks for your help!

  • Sensory perception 
  • Vision 
  • Cognitive science 
  • Factorial design 
  • Perception 
  • Cognition 
  • Covariance 
  • Language 
Sours: https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0102248

Visual and auditory temporal integration in healthy younger and older adults


As people age, they tend to integrate successive visual stimuli over longer intervals than younger adults. It may be expected that temporal integration is affected similarly in other modalities, possibly due to general, age-related cognitive slowing of the brain. However, the previous literature does not provide convincing evidence that this is the case in audition. One hypothesis is that the primacy of time in audition attenuates the degree to which temporal integration in that modality extends over time as a function of age. We sought to settle this issue by comparing visual and auditory temporal integration in younger and older adults directly, achieved by minimizing task differences between modalities. Participants were presented with a visual or an auditory rapid serial presentation task, at 40–100 ms/item. In both tasks, two subsequent targets were to be identified. Critically, these could be perceptually integrated and reported by the participants as such, providing a direct measure of temporal integration. In both tasks, older participants integrated more than younger adults, especially when stimuli were presented across longer time intervals. This difference was more pronounced in vision and only marginally significant in audition. We conclude that temporal integration increases with age in both modalities, but that this change might be slightly less pronounced in audition.


Stimuli that rapidly succeed one after another can be perceived as a single composite stimulus and/or event. When watching a movie, for example, rapid, successive still images are perceived as fluent motion. This is due to a perceptual process named temporal integration, which combines stimuli within an interval up to about 200 ms into an aggregated representation (Hogben & Di Lollo, 1974; Di Lollo, 1980). The duration of the interval varies from person to person, however, and factors that affect cognitive functioning can play a role therein. A person’s age, then, can be an important factor, since aging results in an overall decline or slowing down of the cognitive system (Salthouse, 1996). Yet, how aging affects temporal integration specifically is not yet fully known.

In vision, several studies on temporal integration of visual forms have shown that older adults visually integrate across longer time intervals. For instance, Di Lollo, Arnett, and Kruk (1982) presented participants with two 5 × 5 dot matrices, presented simultaneously side by side, but with the successively plotted dots presented for just 1.5 μs. Participants were asked which of the two matrices contained a missing dot (Di Lollo et al., 1982). To find the missing dot, it is necessary to temporally integrate all dots as if they were presented simultaneously, because consolidating, let alone mentally comparing, 25 positions in such a short time would be impossible. The authors varied the total plotting interval by adjusting the interstimulus interval (ISI) between dots, and found that the younger group needed a shorter plotting interval (60.5 ms) to obtain the same level of 75% task performance as the older group (85 ms). This suggests that the older group temporally integrated the individual sequential dots over a longer interval than the younger group, indicating a longer temporal integration window.

Converging evidence has also been obtained with different tasks, such as color integration (fusion). Kline, Ikeda, and Schieber (1982) briefly presented participants with a green circle followed by a red circle, both presented in the same location. Perceptually, overlaying both circles would result in perceiving a yellow circle. By varying the ISI between the two circles, the authors could measure within what time window participants would temporally integrate the green and red circles and resultantly perceive a yellow circle. The authors found that the older group reported perceiving more color integrations up until the longest ISI, which amounted to a total stimulus duration of 90 ms. The younger group, in contrast, only reported seeing color integrations up to a total stimulus duration of 70 ms. Similarly, in a word recognition task, Kline and Orme-Rogers (1978) measured performance for three-letter words consisting of horizontal and vertical lines, by displaying two halves of random lines of each individual word sequentially. Recognizing the words becomes possible when a participant temporally integrates both halves in a single perceptual representation, which becomes easier when the ISI is small. Across a total stimulus duration range of 100–200 ms, the authors found that the older participants had higher word recognition scores with longer ISIs than the younger participants, which can be explained by a longer temporal integration window for the older group.

As alluded to, one explanation to why aging leads to increased visual temporal integration can be age-related cognitive slowing. According to the processing-speed theory, cognitive slowing would lead to carrying out fewer cognitive operations within a certain timeframe (Salthouse, 1996; Madden & Allen, 2015). When time is limited or processing time is externally constrained, later cognitive operations are then left with less processing time as earlier operations are taking longer to finish. In addition, due to cognitive slowing, memory traces of the results of earlier operations may decay before they can be used for later operations, which illustrates that cognitive slowing causes substantial ‘collateral damage’ apparent as noticeable impairments in daily life activities.

Given the fairly consistent results in the visual domain, one might expect that the auditory modality should be similarly affected. The supposed global nature of cognitive slowing is also compatible with that idea. To wit, measures reflecting other temporal aspects of vision and audition indeed change similarly with age: For both vision and audition, older adults have higher gap detection thresholds (Humes, Busey, Craig, & Kewley-Port, 2009) and are more susceptible to backward masking (Di Lollo et al., 1982; Gehr & Sommers, 1999). However, to our knowledge, there are no studies that have provided direct evidence that the auditory temporal integration window is longer for older adults. In fact, there is indirect evidence pointing to the contrary. An electroencephalographic study on the mismatch negativity (MMN; elicited by a violation in a to-be-expected order or identity of repetitive stimuli; Näätänen, Kujala, & Winkler, 2011) showed that the duration of the auditory temporal integration window does not differ between younger and older adults (Horváth, Czigler, Winkler, & Teder-Sälejärvi, 2007). Using two kinds of oddball experiments (double deviant and stimulus omission), the authors showed that the temporal integration window of their younger participants was around 250 ms, and the window of the older participants was around 200–250 ms.

The lack of evidence for prolonged auditory temporal integration leaves the possibility that aging might be affecting temporal integration differently for each sensory modality. The degree to which integration changes with aging might depend on the relative importance of time in a given sensory modality. In the visual modality, for instance, space is more dominant than time, and it is conceivable that the functionally weakest neurons (i.e., those dealing with temporal aspects) are the first to atrophy when people age. Analogous effects are seen in the body when age-related muscle atrophy is observed (Abate et al., 2007); the so-called “use it or lose it” principle (Schooler, 2007). In perception, the principal dimension of vision is space, but the principal dimension of audition is time (Kubovy, 1988; O’Callaghan, 2008). For example, the borders of visual objects are inherently indicated by coordinates in space, while those of auditory objects are defined in time. In addition, it is easier to imagine an object that is independent of time in the visual domain (e.g., a still image) than in the auditory domain. In line with these conjectures, Humes et al. (2009) showed that auditory gap detection thresholds are lower than the visual ones and that age differences appear to be larger for visual than for auditory stimuli.

Apart from a general effect of time, temporal integration might also be spared more specifically, because temporal integration is required on a daily basis to process and understand speech (Poeppel, 2003; Wallace & Blumstein, 2009): Especially, to analyze vowels, higher level processes map auditory information within 200 ms onto linguistic representations in the form of a phonetic category decision. In addition, even though research showed that older adults have more difficulties with understanding speeded speech (Wingfield, 1996; Gordon-Salant & Fitzgibbons, 2001), Schneider, Daneman, and Murphy (2005) showed that auditory decline and speed-induced stimulus degradation, but not cognitive slowing, may be responsible for lower intelligibility. Thus, it remains conceivable that age-related decline in temporal processing and integration might be lessened in the auditory domain.

Current research

Taken together, there is substantial evidence, indicating that aging increases visual temporal integration, but for the auditory domain, the picture is less clear. Two possibilities exist: first, temporal integration may occur over longer intervals for the older population regardless of the specific sensory modality, which would seem compatible with the notion of general cognitive slowing. Second, differential aging effects on temporal integration in each modality may occur. Such a finding would suggest that the “use it or lose it” principle may apply, meaning that the visual modality could be affected by aging more than in the auditory modality, because the time dimension is less important in vision compared to the space dimension.

The main purpose of the present study was thus to investigate whether aging similarly affects temporal integration in both the visual and auditory domains. Clear evidence from a cross-modality comparison can only be provided with a task that provides a direct measure of temporal integration in each modality equally. In the present study, the visual and auditory tasks were made as similar as possible, using the rapid serial visual presentation (RSVP; Akyürek, Eshuis, Nieuwenstein, Saija, Başkent, & Hommel, 2012) task and its auditory equivalent, rapid serial auditory presentation (RSAP; Saija, Andringa, Başkent, & Akyürek, 2014a). For each task, we tested multiple stimulus durations (40, 70, and 100 ms). If aging affects temporal integration, then this should be reflected in older adults reporting more temporal integration for longer stimulus durations when two targets succeed each other directly (i.e., at Lag 1), in particular. More specifically, the number of temporal integration reports for older adults should decrease at a lower rate with longer stimulus durations compared to younger adults. This should then be reflected in a significant interaction effect of age and stimulus duration.

Experiment 1A: Visual temporal integration



Participants were naive to the purpose of the experiment. Since the experiment relied on visual stimuli, all participants were confirmed to have normal or near-normal vision according to the Ranges of Vision Loss established by the International Council of Ophthalmology (2002). The participants’ visual acuity was measured (with lenses or glasses if required) using the Landolt C test. The mean visual acuity for the young group was LogMAR −0.16 and for the older group LogMAR −0.02. Figure 1 shows the visual acuity as a function of age. Furthermore, mental flexibility and normal cognitive functioning were confirmed with the Trail-Making Test Parts A and B (Chanmugam, Triplett, & Kelen, 2013). Three older adults were excluded from participation, because one was suffering from macula pucker, one was unable to perform the task, and one had a stroke in the past. After exclusion, 19 young students of the University of Groningen (6 male and 13 female) with a mean age of 20 years (ranging from 17 to 23 years) and 19 older adults (16 male and 3 female) with a mean age of 70 years (from 65 to 81 years) participated in the study. Younger participants received course credit or monetary compensation, while older participants only received monetary compensation. Informed consent was obtained in writing before participation, and the study was approved by the Ethical Committee of the Department of Psychology at the University of Groningen.

Experiment 1A This graph shows the visual acuity in LogMAR for young and older participants by age

Full size image

Apparatus and stimuli

The experiment was implemented with E-Prime Professional (Psychology Software Tools, Pittsburgh, PA) running on a desktop computer with Microsoft Windows XP. The visual stimuli were presented on a 19-inch CRT screen, which refreshed at 100 Hz with a resolution of 1024 × 768 pixels in 16-bit color, and which was placed at a viewing distance of approximately 60 cm. The participants’ responses were collected via a keyboard.

The target stimuli consisted of the symbols / \ o and their combinations, as shown in Fig. 2. They were at most 49 pixels in height and 33 pixels in width (approximately 1.6° and 1.1° of visual angle, respectively) and were displayed in red (RGB 255, 0, 0; 91 cd/m2). The targets were chosen, such that their features did not overlap with each other (e.g., / was never presented with the X). The distractor stimuli were drawn without replacement from the modern Latin alphabet (excluding I, J, K, L, O, and X to avoid confusion with the target symbols). The distractor stimuli, as well as the fixation cross, were all printed in bold 52 pt. Courier New font and colored in black (RGB 0, 0, 0; 2 cd/m2). The targets and distractors were about equal in size. The background color was always light gray (RGB 192, 192, 192; 265 cd/m2).

Experiment 1A Example of a typical trial to illustrate the procedure and visual stimuli. The empty boxes with solid lines represent blank periods of 100 ms. The empty boxes with dashed lines represent the succession of multiple distractor stimuli (i.e., black letters). The target stimuli were always presented in red. For each trial, all stimuli were of equal duration and were presented for 40, 70, or 100 ms. Each stimulus was separated by an ISI of 10 ms

Full size image


The experiment consisted of a short block of practice trials and continued with 496 experimental trials with an optional break halfway, lasting for approximately 60–90 min. At 100 ms, after a trial was initiated by a participant, the fixation cross was displayed for 200 ms. Then, 19 stimuli succeeded each other, all of which were on screen for 40, 70, or 100 ms and followed by a 10 ms blank screen each (50, 80, and 110 ms SOA, respectively; 1/3 of trials each). On 94.4% of the trials, two of these stimuli were targets (T1 and T2), while the others were distractors. T1 appeared as either the fifth item or the seventh item in the stream and T2 followed T1 with either 0, 2, or 7 distractors in-between, referred to as Lag 1, 3, or 8 (31.5% of trials each). On 5.6% of the trials, T1 was a solo target.

The participants were told that each trial could contain one or two targets, and they were asked to identify each of them. After each stream, a 100 ms blank screen was presented, after which the participants were asked to enter the identity of T1 and then that of T2 on the numerical keypad. Each target response alternative was labeled on the numerical keypad. If a target was not spotted, then an empty response could be given by pressing the Enter key. However, guessing was encouraged when a participant was unsure about the identity of a target. Figure 2 shows an example of a trial that illustrates the procedure.


Of main interest were the reports of integrated percepts (i.e., reporting the integrated percept of the combined features of T1 and T2) that were reported as a single response (i.e., no second response was entered). These responses were regarded as strict integrations, and indicated that the observer only perceived a single target, which constituted of the integrated combination of T1 and T2. Second, task performance was analyzed, which reflects correct response accuracy of the target identities and their temporal order. Analyses were performed on the number of trials in which T1 was correctly reported, and in which T2 was correctly reported given that T1 was correct as well (T2|T1). T1 was also considered correctly reported when the integrated percept of T1 and T2 was reported (as was T2|T1).

The data were in the form of count data, and because the variance of the data for each analysis was larger than the data’s mean, all data best fitted the negative binomial distribution. Therefore, the data were analyzed using generalized estimating equations using a negative binomial distribution with log link. For each analysis separately, the overdispersion parameter (α) was estimated and the working correlation matrix (WCM) was chosen based on the best goodness of fit [i.e., lowest quasi-likelihood under the independence model criterion (QIC); Pan, 2001]. Each analysis included the two within-subject variables’ stimulus duration (40, 70, and 100 ms) and T1–T2 lag (1, 3, and 8), as well as the between-subject variable age group (young and older participants). Strict integrations were expected to happen mostly at Lag 1 due to the short distance between targets and the lack of distractors in-between, and therefore, additional analyses were performed on the data of Lag 1 only, whereby T1–T2 lag was removed as a variable. For each test, a significance level of 0.05 was used.

The strict integration reports were represented as relative frequencies, that is, relative to all trials in which both target identities are retained regardless of their positions (i.e., strict integrations, order reversals, and both correct responses). For reference, the Appendix contains figures with the absolute integration rates for all experiments reported here. To account for this relativity, the offset for each combination of subject and condition was included in these analyses and was calculated as the natural log of the exposure (i.e., of the number of trials that include strict integrations, order reversals, and both correct responses, per subject and condition). For the (T2|T1) accuracy, the offset for each combination of subject and condition was calculated as the natural log of the exposure of the number of trials in which T1 was correct. For T1 accuracy, there was no relativity, so for each subject and condition, all trials could be included. Therefore, the T1 offset for all conditions and subjects was set to the natural log of the total number of trials per condition and subject [ln(52) ≈ 3.95].

The estimated marginal means of the analyses of relative frequencies of strict integration reports were plotted in bar graphs. The estimated marginal means of the T1 and (T2|T1) accuracies were also plotted in bar graphs, together with the accuracies when report order is ignored (e.g., when T1’s identity is correct regardless of T1’s position, namely, including T1 reported as T2, order reversals, and strict integrations).


A full factorial analysis (WCM = autoregressive, α = 15.322) was performed on the relative frequencies of strict integration (i.e., relative to strict integrations, order reversals, and both correct responses), which are shown in Fig. 3. The frequency of strict integrations was significantly affected by lag, χ2(2, N = 342) = 64.7, p < 0.001, by stimulus duration, χ2(2, N = 342) = 95.5, p < 0.001, and by their interaction lag*duration, χ2(4, N = 342) = 55.5, p < 0.001. Figure 3 shows that reports of strict integrations are most prominent at Lag 1 and become less frequent with longer lags and longer stimulus durations. Strict integrations were also affected by group, χ2(1, N = 342) = 19.6, p < 0.001, as well as by the interactions of group*lag, χ2(2, N = 342) = 8.8, p < 0.015, and group*duration, χ2(2, N = 342) = 21.8, p < 0.001.

Experiment 1A This graph shows the estimated marginal means of the analyses of relative frequency of strict integrations for all combinations of stimulus duration, lag, and age group, as a percentage of the total number of trials in which both target identities were preserved. Error bars represent ±1 standard error of the mean

Full size image

An additional analysis for Lag 1 only (WCM = unstructured, α = 3.029) showed that stimulus duration was a significant factor, χ2(2, N = 114) = 68.9, p < 0.001, which indicates that shorter stimulus durations resulted in more reports of strict integrations. Even more importantly, older adults were more influenced by stimulus duration than young adults, revealed by an interaction effect of group*duration, χ2(2, N = 114) = 26.7, p < 0.001, meaning that older adults integrated more often than young adults at longer stimulus durations. In addition, older adults reported more strict integrations at Lag 1 over all three durations, χ2(1, N = 114) = 17.6, p < 0.001. These effects can be seen in more detail, as shown in Fig. 3.

Another factorial analysis (WCM = unstructured, α = 2.015) was performed on the frequency of trials, where T1 was correct. The average accuracy of T1 per group for each lag and stimulus duration are shown in Fig. 4, together with the average accuracy when report order is ignored (i.e., relaxed criterion). T1 accuracy was significantly affected by lag, χ2(2, N = 342) = 213.3, p < 0.001, and stimulus duration, χ2(2, N = 342) = 137.8, p < 0.001, as well as by their interaction lag*duration, χ2(4, N = 342) = 43.3, p < 0.001. Figure 4 reveals that T1 accuracy was higher for each stimulus duration when lags were longer, as well as for each lag when the stimulus durations were longer. The accuracy of T1 also differed per age group, χ2(1, N = 342) = 21.4, p < 0.001, indicating that the younger group overall had higher performance. In addition, group*lag was significant, χ2(2, N = 342) = 8.6, p < 0.015, as well as group*lag*duration, χ2(4, N = 342) = 9.9, p < 0.045.

Experiment 1A The solid bars at the front show the estimated marginal means of the analyses on T1 task performance in percent correct, plotted for all combinations of stimulus duration and lag, for both age groups. The transparent bars at the back show the same analyses if report order is ignored (i.e., relaxed accuracy criterion). Error bars represent ±1 standard error of the mean

Full size image

A final full factorial analysis (WCM = independent, α = 2.667) was performed on the number of trials, where T2 was correct, given that T1 was correct as well (T2|T1). Figure 5 shows the average accuracy of T2|T1 per group and for each lag and stimulus duration, as well as the average accuracy when report order is ignored. T2|T1 accuracy was significantly affected by lag, χ2(2, N = 342) = 108.5, p < 0.001, and stimulus duration, χ2(2, N = 342) = 138.8, p < 0.001, as well as by their interaction lag*duration, χ2(4, N = 342) = 36.6, p < 0.001. Figure 5 reveals that T2|T1 accuracy was higher for each longer lag or longer stimulus duration. The accuracy of T2|T1 also differed per age group, χ2(1, N = 342) = 22.1, p < 0.001, indicating that the younger group overall performed better. In addition, group*lag was significant, χ2(2, N = 342) = 9.7, p < 0.01, as well as group*duration, χ2(2, N = 342) = 9, p < 0.015, and group*lag*duration, χ2(4, N = 342) = 14.7, p < 0.01.

Experiment 1A The solid bars at the front show the estimated marginal means of the analyses on T2|T1 task performance in percent correct, plotted for all combinations of stimulus duration and lag, for both age groups. The transparent bars at the back show the same analyses if report order is ignored. Error bars represent ±1 standard error of the mean

Full size image

Summarizing, older adults showed more integration than younger adults for visual stimuli, particularly for the longer stimulus durations tested. Elevated integration frequency was even observed at Lag 3, when 40 ms stimulus duration was used, for the older adults. For them, the speed of presentation seemed to overcome the inhibitory effects on integration of the intervening distractors. The younger group rarely integrated at Lag 3, even at the fastest presentation speeds. General task performance of the older adults, as measured by both T1 and T2|T1 accuracies, was also lower than that of the younger adults. Overall, the results were thus in line with expectations.

Experiment 1B: The effect of retinal illuminance on visual temporal integration

To be able to interpret the results of Experiment 1A unambiguously, it is necessary to exclude the possibility that the observed age-related differences could be due to purely sensory factors, such as increasing opacity of the lens with age. Specifically, it is conceivable that older people integrates more, because their retinal illuminance is reduced (Coltheart, 1980; Di Lollo, Hogben, & Dixon, 1994). Older people have on average a reduction of around a 0.5 log unit of retinal illuminance compared to that of younger people (Weale, 1963). To investigate whether the older adults in Experiment 1A perceived more integrated stimuli because of an inverse intensity effect (i.e., more integration with dimmer stimuli), a new group of younger adults was tested with 34% screen brightness instead of 100% in Experiment 1B, which simulates an approximate 0.5 log unit reduction in retinal illuminance. The experiment was otherwise identical to Experiment 1A (young group only).



Twenty-three young students of the University of Groningen (20 male and 3 female) with a mean age of 20 years (from 17 to 34 years) participated. All participants had normal or near-normal vision: the mean visual acuity for this new group of young adults was LogMAR −0.14. Figure 6 shows visual acuity as a function of age. All participants received course credit for their participation.

Experiment 1B This graph shows the visual acuity in LogMAR for the participants as a function of age

Full size image

Apparatus and stimuli

The only difference with Experiment 1A was that the brightness of the screen was set to 34% instead of 100% (in hardware), simulating reduced retinal illuminance as might be experienced by older observers. The red target stimuli were now displayed at 39 cd/m2 and the light gray background at 109 cd/m2.


The analyses were focused on relative frequencies of strict integration reports. First, we tested whether the reduced brightness in Experiment 1B resulted in more strict integration reports than in Experiment 1A; therefore, the main analysis included the between-subject variable group (comparing the young participants from Experiment 1A with those from Experiment 1B) and the two within-subject variables’ stimulus duration (40, 70, and 100 ms) and T1–T2 lag (1, 3, and 8). Second, a detailed analysis was performed on Lag 1 with the within-subject variable stimulus duration, for both the young participants of Experiments 1A and 1B.


A full factorial analysis (WCM = autoregressive, α = 24.360) was performed on the relative frequencies of strict integration, which are shown in Fig. 7. The frequency of strict integrations was significantly affected by lag, χ2(2, N = 378) = 100.387, p < 0.001, by stimulus duration, χ2(2, N = 378) = 30.09, p < 0.001, and by their interaction lag*duration, χ2(3, N = 378) = 19.41, p < 0.001. Figure 7 shows that reports of strict integrations were most prominent at Lag 1 and became less frequent with longer lags and longer stimulus durations, as observed previously. Strict integrations were also affected by the interaction of group*duration, χ2(2, N = 378) = 7.82, p < 0.025, and the interaction of group*lag*duration, χ2(2, N = 378) = 7.05, p < 0.035, reflecting that low luminance seemed to decrease integration frequency in some conditions only, particularly at Lag 1, and at 40 ms duration.

Experiment 1B This graph shows the estimated marginal means of the analyses of relative frequency of strict integrations for all combinations of stimulus duration, and lag, as a percentage of the total number of trials in which both target identities were preserved. The data from the young group of Experiment 1A (full luminance) are re-plotted next to the low luminance group for reference. Error bars represent ±1 standard error of the mean

Full size image

An additional analysis for Lag 1 only (WCM = unstructured, α = 31.216) showed that only stimulus duration was a significant factor here, χ2(2, N = 126) = 101.29, p < 0.001. The lack of a significant group factor indicates that luminance did not have a significant effect on strict integration reports.

Even though Experiment 1B could not perfectly match the retinal illuminance of older observers (e.g., due to constant room lighting), the reduction in screen luminance was substantial enough that a sensory-driven rise in integration should have been revealed. However, the findings did not at all support the idea that reduced retinal illuminance might have fostered integration in the current task. As shown in Fig. 7, there was actually a trend in the opposite direction: Reduced brightness resulted in the perception of fewer integrated stimuli. Therefore, we can conclude that older people do not temporally integrate more, because they perceive less brightness. The nature of the present task, in which dark stimuli appear on a light background (i.e., with inverse contrast), might have played a mediating role therein. In addition, it is conceivable that a reduced ability to perceive darker targets may actually have limited the opportunity to integrate, as integration requires at least the perception of the stimulus features.

Experiment 2: auditory temporal integration

The auditory Experiment 2 was carried out after Experiment 1, its visual counterpart, produced the expected pattern of results. It was similar to the RSAP experiment described in Saija et al. (2014a) but with two additional stimulus durations (40 and 70 ms). Similar to the RSVP experiments, during the RSAP experiment, a participant was presented with a stream of auditory instead of visual targets and distractors. The participant then had to report which targets were heard. The two auditory targets consisted of complex tones, which could be integrated pairwise into two-formant synthetic vowels, analogous to the visual target combinations that were enabled in the RSVP experiments. During a pilot study with older participants, it became clear that they were unable to discriminate between the original target stimuli and remember them, maybe as a result of age-related changes in temporal fine structure processing (Füllgrabe, 2013), age-related short-term memory deficits (Chen & Naveh-Benjamin, 2012), or some loss of auditory acuity (even if within the range of normal hearing; Martini & Mazzoli, 1999). Therefore, the stimuli were modified in such a way that the older participants could discriminate the target stimuli more easily (as detailed below).



Participants were naive to the purpose of the experiment. Since the experiment relied on auditory stimuli, all participants were selected to have normal or near-normal hearing. They reported to have normal hearing, and their audiometric thresholds were tested using the definition of normal hearing from Martini and Mazzoli (1999), namely, that the four-tone pure average across 0.5, 1, 2, and 4 kHz should be 20 dB HL or lower. Figure 8 shows the audiometric thresholds for each individual for both age groups. In addition, all participants were required to take the Trail-Making Test Parts A and B to test for mental flexibility and normal cognitive functioning (Chanmugam et al., 2013). An additional requirement was to be a fluent speaker of Dutch, as the stimuli were based on Dutch vowels. Two young and seven older participants were excluded from participation, because they found the training too difficult. In addition, eight older participants were excluded due to insufficient hearing, and two were excluded, because they were unable to successfully finish the Trail-Making Test Part B. After exclusion, 22 young students of the University of Groningen (11 male and 12 female) with a mean age of 20 (from 18 to 26) participated in the experiment for course credit. In addition, 22 older adults (7 male and 16 female) with a mean age of 65 (from 60 to 71) participated for monetary compensation. Informed consent was obtained in writing before participation, and the study was again approved beforehand by the Ethical Committee of the Department of Psychology at the University of Groningen.

Experiment 2 This graph shows the auditory acuity for the young and older participants in dB hearing level per frequency, plotted for the ear with the lowest hearing levels for each participant

Full size image

Apparatus and stimuli

The experiment was implemented in Matlab (; R2015a) using Psychtoolbox (3.0.12; Brainard, 1997; Pelli, 1997; Kleiner et al., 2007) running on a Mac Pro with Mac OS X (10.10.4). Auditory stimuli were presented diotically through a Sennheiser HD 600 headphone, connected to an Echo Audiofire 4 external soundcard and a Lavry Engineering DA10 digital-to-analog converter. Responses were collected with a standard keyboard. Participants were tested in a sound-isolated booth.

The stimuli were created in Praat using a Klattgrid (Weenink, 2009), which is a speech synthesizer based on the Klatt synthesizer (Klatt, 1980; Klatt & Klatt, 1990). The Klattgrid program was used to create three Dutch vowels/a/,/i/and/ø/ (Pols, Tromp, & Plomp, 1973) with a pitch tier of 120 Hz, as well as the distractor tone, which was always the same and repeated during the experiment. Each vowel consisted of the first four formants (F1–F4; see Table 1). The use of four formants instead of two as in Saija et al. (2014a) ensured that the artificial vowels sounded more rich and more similar to natural vowels, making them easier to recognize and to discriminate between them. Each artificial vowel was divided into two parts, and each part was a possible target sound. One part contained F1 and F3, and was perceived as being lower in timbre than the distractor, because most energy was at F1. The other part contained F2 and F4, and was perceived as being higher in timbre as most energy was at F2. F1 was lower in frequency than the distractor and F2 was higher (see Table 1). The bandwidth of F1 was set to 50 Hz, and the bandwidth of each subsequent formant was enlarged by 50 Hz compared to the previous formant. Part 1 was set at 65 dB SPL and each second part was set at a lower intensity (see Table 1) that would result in the best perception of the artificial vowel when both parts are combined. In addition, a ramp of 5 ms was placed at each on- and offset to prevent audible distortions of potential spectral splatter. The three bottom panels of Fig. 9 show spectrograms of the three vowels.

Full size table

Experiment 2 The top panel shows the spectrogram (window length = 5 ms; dynamic range = 70 dB) of a typical Lag 1 trial. From left to right, a number of distractor stimuli are presented, followed by part 1 (F1 + F3) of the vowel/ø/, and then part 2 (F2 + F4) of the same vowel, followed by more distractor stimuli. Stimuli were 40, 70, or 100 ms in duration, and were always separated by a 10 ms silent gap. The three bottom panels show spectrograms (window length = 5 ms; dynamic range = 45 dB) of the three 4-formant vowels/a/,/i/and/ø/, in this example with a duration of 70 ms

Full size image


The participants were asked to classify the targets as one of five response alternatives; the three different vowels, a tone that was lower in timbre than the distractor, or a tone that was higher than the distractor. All response alternatives were labeled on the numerical keyboard.

First, participants had to be trained to be able to identify all different targets. Therefore, they were given a few minutes to listen to each target (embedded in a short series of distractors) as often as they wanted until they felt acquainted with the targets. After that, they were given a short training session in which they were presented with the targets one by one. They then had to indicate which target they thought was presented, and they received visual feedback, together with an auditory presentation of the target they responded with and the presented target. Once the participants were able to distinguish the targets, a short final practice session followed consisting of a number of practice trials, which were similar to those in the experiment. Afterwards, the actual experiment started and consisted of 513 trials. A trial consisted of a series of 18 sequential stimuli, from which one or two could be targets and the rest distractors. On 94.74% of all trials, two targets were presented, in which both targets should belong to the same formant pair (i.e., T1 as F1 and T2 as F2, or vice versa). T1 appeared as the fifth or seventh stimulus, and T2 appeared at Lag 1, 3, or 8 (each 31.58% of all trials). On 5.26% of all trials, T1 was the single target, in which it could be a vowel (1.75%) or single formant (low tones 1.75%; high tones 1.75%). Stimuli had durations of 40, 70, or 100 ms (1/3 of all trials each), and were separated by a 10 ms gap. The top panel of Fig. 9 shows a spectrogram of a part of a typical Lag 1 trial.

The participants started a trial by pressing the spacebar. After each stream of stimuli, the participants entered what they heard as the first and second targets in their perceived order. When participants only heard a single target, they were able to give an empty response as the second target by pressing the Enter key. The experiment, including the training session, lasted approximately 1.5 h for the younger adults and 2 h for the older adults.

Data analysis

To classify a single response as a strict integration, the response should be the vowel that would have been the product of the combination of both targets. For example, if a participant reported to have only heard the/a/and no other target, and T1 was the F1 + F3 of/a/and T2 the F2 + F4 of/a/(or vice versa), then this report would be classified as a strict integration. Otherwise, the data analysis was similar to that of Experiment 1, except that the offset for T1 accuracy was ln(54) ≈ 3.99.


A full factorial analysis (WCM = exchangeable, α = 23.830) was performed on the relative frequencies of strict integration, which are shown in Fig. 10. The frequency of strict integrations was significantly affected by lag, χ2(2, N = 396) = 154.9, p < 0.001, by stimulus duration, χ2(2, N = 396) = 51, p < 0.001, and by their interaction lag*duration, χ2(4, N = 396) = 32.4, p < 0.001. As shown in Fig. 10, strict integrations were most frequent at Lag 1, and their frequency decreased with longer lags and longer stimulus durations. Strict integrations were also affected by group, χ2(1, N = 396) = 5.3, p < 0.025, as well as by the interaction of group*lag*duration, χ2(3 N = 396) = 9.3, p < 0.03.

Experiment 2 This graph shows the estimated marginal means of the analyses of relative frequency of strict integrations for all combinations of stimulus duration, lag, and age, as a percentage of the total number of trials in which both target identities were preserved. Error bars represent ±1 standard error of the mean

Full size image

An additional analysis for Lag 1 only (WCM = autoregressive, α = 2.56) showed that stimulus duration was a significant factor, χ2(2, N = 132) 15.3, p < 0.001, which indicates that shorter stimulus durations resulted in more reports of strict integrations. In addition, older adults marginally reported more strict integrations at Lag 1 over all three durations, χ2(1, N = 132) = 3, p = 0.085. These effects can be seen in more detail, as shown in Fig. 10.

Another full factorial analysis (WCM = unstructured, α = 3.587) was performed on T1 accuracy, as shown in Fig. 11. T1 accuracy was significantly affected by lag, χ2(2, N = 396) = 74.2, p < 0.001, and stimulus duration, χ2(2, N = 396) = 34.5, p < 0.001, as well as by their interaction lag*duration, χ2(4, N = 396) = 86.5, p < 0.001. Figure 11 reveals that T1 accuracy was higher for each stimulus duration at longer lags, as well as for each lag when the stimulus durations were longer. The accuracy of T1 also differed per age group, χ2(1, N = 396) = 5.6, p < 0.02, indicating that the younger group performed better overall.

Experiment 2 The solid bars at the front show the estimated marginal means of the analyses on T1 task performance in percent correct, plotted for all combinations of stimulus duration, lag, and age group. The transparent bars at the back show the same analyses if report order is ignored. Error bars represent ± 1 standard error of the mean

Full size image

The last full factorial analysis (WCM = autoregressive, α = 3.757) was performed on T2|T1 accuracy, as shown in Fig. 1. T2|T1 accuracy was significantly affected by lag, χ2(2, N = 396) = 17.5, p < 0.001, and lag*duration, χ2(4, N = 396) = 13.6, p < 0.01. Figure 12 reveals that T2|T1 accuracy was higher for each stimulus duration when lags were longer, as well as for each lag when the stimulus durations were longer (except for Lag 3 and 8 from 70 to 100 ms). The accuracy of T2|T1 also differed per age group, χ2(1, N = 396) = 10.1, p < 0.002, indicating that the younger adults were also better able to identify the second target.

Experiment 2 The solid bars at the front show the estimated marginal means of the analyses on T2|T1 task performance in percent correct, plotted for all combinations of stimulus duration, lag, and age group. The transparent bars at the back show the same analyses if report order is ignored. Error bars represent ±1 standard error of the mean

Full size image

Comparison of Experiment 1A and Experiment 2

To analyze whether temporal integration in both rapid serial presentation experiments occurred similarly, we performed a GEE test with experiment, age group, and stimulus duration as factors, on the strict integration data for Lag 1 only (WCM = unstructured; α = 6.807). The test revealed that experiment, χ2(1, N = 246) = 5.4, p < 0.025, age, χ2(1, N = 246) = 13.2, p < 0.001, and duration, χ2(2, N = 246) = 64.4, p < 0.001, was significant main factors, as expected. The significant interaction effects were experiment*duration χ2(2, N = 246) = 6.5, p < 0.04 and age*duration χ2(2, N = 246) = 16.7, p < 0.001. The significant effect of experiment indicates that temporal integration was more frequent in the visual domain, as is evident from comparing Figs. 3 and 10. The interaction effect of experiment and duration indicated that integration decreased more sharply as duration increased in the visual modality. The interaction between age and duration showed that overall, this decrease was attenuated for the older participants; they integrated comparatively more at the longer durations. However, the absence of interaction effects of experiment*age and experiment*age*duration indicates that age did not influence temporal integration differently per experiment. This means that aging affected temporal integration similarly in both modalities, even if it appeared from the individual analysis of Experiment 2 to do so less strongly in audition.

General discussion

Previous literature provided evidence that aging results in more temporal integration in vision; however, evidence for the auditory domain remained inconclusive (e.g., Horváth et al., 2007). Therefore, the primary aim of this study was to investigate if aging affects temporal integration similarly in the visual and auditory domains. To this end, we conducted two rapid serial presentation experiments, visual and auditory, aiming to obtain a direct, comparable measure of temporal integration in each modality.

The results of the visual task (Experiments 1A and 1B) showed that temporal integration was significantly affected by aging at Lag 1. The older adults reported more temporal integration overall than the younger adults did. Most importantly, the interaction effect of age and stimulus duration at Lag 1 (where both targets succeeded each other directly) was significant. This showed that for older adults, visual temporal integration decreased less steeply with increasing stimulus duration, which means that the older adults integrated more at longer stimulus durations, as would be expected for a longer temporal window of integration. The results of the auditory experiment, however, showed a weaker aging effect on temporal integration: older adults reported only marginally more temporal integrations at Lag 1 than younger adults. In addition, there was no significant interaction effect between age and duration at Lag 1. Yet, the analysis of temporal integration at Lag 1 between both experiments revealed that the general pattern of performance was not reliably different. In other words, age influenced temporal integration similarly, even if temporal integration was most apparent for the visual modality as indicated by a significant main effect of experiment (cf. Figs. 3, 10). From these facts combined, we can conclude that aging does affect temporal integration in both the visual and auditory domains, but that the effect may be somewhat weaker in the latter.

Locus of age-related differences in temporal integration

In the current experiments, we aimed to minimize the differences in visual and auditory sensory properties between the age groups, so that any differences in results could be attributed to differences in cognitive rather than perceptual capabilities (cf. Lindenberger & Baltes, 1994). Because it is not feasible to fully remove all sensory differences between the age groups, we aimed to reduce the differences to a minimum using participants that had normal vision and hearing according to the respective standards (International Council of Ophthalmology, 2002; Martini & Mazzoli, 1999). It must nonetheless be acknowledged that small sensory differences between the groups did remain, which might have contributed to differences in temporal integration. However, the results of Experiment 1B suggested that such a sensory effect can be discounted, since the data showed a pattern contrary to what would be expected if sensory degradation caused the age-related differences in temporal integration: we found less rather than more integration with reduced illuminance.

It, therefore, seems more likely that the differences in temporal integration originate from a more upstream locus in the perceptual processing pathway. For instance, older people have decreased early sensory memory abilities for short, individual stimuli (Fogerty, Humes, & Busey, 2016), making it harder to successfully keep fine-grained, individual stimuli in store. This might result in temporally blurred representations due to longer integration windows. Older people also seem to have more difficulties with separating and encoding short, individual, sequential stimuli because of decreased temporal processing capabilities, which might again result in overlapping representations. Supporting evidence has been obtained from gap detection tasks (Di Lollo et al., 1982; Humes et al., 2009), in which younger people can detect smaller gaps, and from temporal order judgments tasks, in which older people need longer ISI and stimulus durations to successfully judge the order of two sequential visual or auditory stimuli (Kolodziejczyk and Szelag, 2008; Ulbrich, Churan, Fink, & Wittmann, 2009).

Indeed, by themselves, such more function-specific theories are already quite capable of explaining why older people may have longer temporal windows and integrate more than younger people. However, it may be noted that the concept of cognitive slowing arguably encompasses these more specific theories. To recap, the processing-speed theory states that cognitive slowing leads to degraded cognitive functioning (Salthouse, 1996; Madden & Allen, 2015), which impacts perception according to the common cause principle. An individual with slower cognitive processing speed can carry out fewer cognitive operations within a certain timeframe (i.e., decreased temporal processing capabilities). Consequently, with limited processing time, subsequent cognitive operations are left with less processing time as earlier operations are taking longer to finish. Because of this, memory traces of the results of earlier operations may decay or become less strong, which make them susceptible for merging with subsequent memory traces. It, therefore, seems most parsimonious to refer more generally to cognitive slowing as the underlying mechanism that affects temporal integration with aging, regardless of the modality.

Although a general theory for the presently observed effects is appealing, the current data leave the possibility that the prominence of time in audition can at least slightly weaken the age-related differences in that modality. However, not all alternative explanations for this slight discrepancy between modalities can be ruled out. Because sensory and cognitive aging may correlate (e.g., Humes, Busey, Craig, and Kewley-Port, 2013; Roberts & Allen, 2016; Wayne & Johnsrude, 2015), the strict exclusion criteria applied out of necessity in Experiment 2 may have resulted in a relatively high-performing sample, which may have translated into comparatively modest integration rates. Thereby, the age-related effect may have become more difficult to detect. Another possibility is that the weaker effect in audition was due to the nature of the stimuli. One might suppose that the targets in the visual experiment were less meaningful than those in the auditory experiment (i.e., vowels) and that this difference could have mediated the integration process, such that auditory targets were less integrated. This account nevertheless seems problematic, because (1) not all auditory targets were meaningful vowels, (2) integrated reports could only consist of vowels combined from complex tones, which means that an increase in reports of integrated vowels should be expected, and (3) the symbols used in the visual experiment might also be regarded as meaningful (consider, for instance, the target “X”).

Relation to neurophysiology and attentional blink

In neurophysiological terms, age-related cognitive decline is associated with myelin loss in the white matter of brain regions that myelinate late during brain development (Lu et al., 2011; Salami, Eriksson, Nilsson, & Nyberg, 2012; Lu et al., 2013), such as the prefrontal cortex (often associated with executive functioning, memory and attention) and the genu of the corpus callosum, which connects the prefrontal cortex on both hemispheres (Bloom & Hynd, 2005). Because the axons in these regions are less thickly myelinated, they are more fragile and sensitive to age-related degradation. In turn, such degradation diminishes the myelin’s function to accelerate transmission speed of action potentials through leaping conduction, which could possibly lead to cognitive slowing. Because the prefrontal cortex is related to attention and working memory, a general account of cognitive slowing thus fits our results quite well. Namely, in the currently used rapid serial presentation tasks, subjects need a sufficient level of attention and working memory capacity to successfully detect, identify, and remember the rapidly presented targets while ignoring intermediate distractors.

Furthermore, according to the simultaneous-type serial token model (Bowman & Wyble, 2007), two targets can be combined into a single target representation or episodic memory trace when the temporal overlap between the activation of both targets is adequate. Perceptually combining two targets in such a way costs less mental effort, as was shown by Wolff, Scholz, Akyürek, and van Rijn (2015), meaning that working memory is burdened less. Because older adults generally struggle more on attentional and cognitive tasks (Craik & Salthouse, 2011), it is conceivable that they use this temporal integration mechanism more frequently, as it may serve as a compensation mechanism to save mental resources. Most compensation mechanisms that are used by older adults result in increased brain activity compared to younger adults, to compensate for the age-related changes in the brain (Grady, 2012). In our tasks, to successfully detect, identify, remember, and keep up with the rapidly presented targets and ignore distractors, it is conceivable that older adults integrate more, because they have less mental resources or neuronal connections to perform this demanding task.

If so, it might be hypothesized that the brain activity of older adults in the prefrontal cortex increases as a way to keep up with the fast pace, resulting in an attempt to increase attention to the targets. Previous research showed that if more attention is given to targets temporal integration also increases (Visser & Enns, 2001), and also that successful temporal integration is related to increased amplitudes of the N1, N2, and late P3, which are event-related potential components related to attention (Akyürek, Schubö, & Hommel, 2010). Note that even though temporal integration might come with increased brain activity, it is nonetheless less demanding (or costs less mental effort) than keeping up with each single stimulus at a time, making it a suitable compensation mechanism for older adults with fewer neuronal connections (and thereby likely fewer mental resources) to begin with. In practice, such compensation would result in a prolonged temporal integration window, as longer periods are covered in a single episodic memory trace, which can be seen in our experiments, where the older adults integrated more at longer durations.

These interpretations fit well with the previous attentional blink (AB) results. The AB is expressed in the difficulty of perceiving the second of two targets (typically in RSVP) if it arrives between 150 and 500 ms after the first (Raymond, Shapiro, & Arnell, 1992). Importantly, recent work on individual differences suggests that people with a larger AB tend to integrate more (Willems, Saija, Akyürek, & Martens, 2016), which is in line with task performance in terms of effective allocation of cognitive resources, as given above. Furthermore, the previous research has also shown that the AB is larger for older adults in both modalities (Lahar, Isaak, & McArthur, 2001; Slawinski & Goddard, 2001). The current results show a similar pattern, both for integration, as discussed, and for target accuracy also: age had a significant effect on T1 and T2|T1 accuracies in both modalities, meaning that for both measures, older adults had lower accuracy over all conditions. One caveat is that even though we controlled for normal visual and auditory acuity, in practice, the acuity was on average slightly better for the younger groups, which might have contributed to the differences in accuracy.

Finally, a further advantage of a prolonged temporal integration window, besides the reduction of mental effort, is that it might be beneficial for high-level compensatory mechanisms for better perception of degraded speech, such as measured in studies of the phonemic restoration effect (Warren, 1970; Başkent, 2012). With phonemic restoration, listeners are able to restore degraded speech that contains missing speech segments that are filled by loud noise, using top–down knowledge to fill in the missing segments and combine the available and filled-in loose segments into coherent understandable speech. Saija, Akyürek, Andringa, and Başkent (2014b) showed that older adults, in some conditions, have a larger restoration effect than younger adults, and concluded that this might be due to the older adults’ superior language skills, vocabulary, and world knowledge. However, in light of the current results, it might be that temporal integration plays a role as well. Namely, Fig. 10 shows that with auditory stimuli, older adults integrated more at Lag 3 than younger adults (most prominent at 40 ms stimulus duration, and similar to the visual task). Normally, temporal integration would occur when two targets are presented in succession without intermediate distractors. However, for the older adults in this case, integration also happened with intermediate distractors at Lag 3. Such integration of two targets spanning over two intermediate distractors is not seen with young adults. With phonemic restoration, listeners also have to combine information of speech segments that are separated or masked by intermediate noise. Therefore, it is conceivable that a prolonged temporal integration window, as is seen with older adults, might have a positive effect on the phonemic restoration ability.


In summary, the current results show that the older adults integrated overall more than the young adults, independent of modality. The older adults also integrated comparatively more at longer durations than the young adults. This effect was most clearly observed in the visual domain, and seemed less pronounced in audition. These results seem to reflect a general, cognitive–perceptual change with age, with the tentative addition that the prominence of time in audition may weaken this effect for auditory temporal integration.


  1. Abate, M., Di Iorio, A., Di Renzo, D., Paganelli, R., Saggini, R., & Abate, G. (2007). Frailty in the elderly: The physical dimension. Europa Medicophysica,43(3), 407–415.

    PubMed Google Scholar

  2. Akyürek, E. G., Eshuis, S. A., Nieuwenstein, M. R., Saija, J. D., Başkent, D., & Hommel, B. (2012). Temporal target integration underlies performance at lag 1 in the attentional blink. Journal of Experimental Psychology: Human Perception and Performance,38(6), 1448–1464. doi:10.1037/a0027610.

    ArticlePubMed Google Scholar

  3. Akyürek, E. G., Schubö, A., & Hommel, B. (2010). Fast temporal event integration in the visual domain demonstrated by event-related potentials. Psychophysiology,47(3), 512–522. doi:10.1111/j.1469-8986.2010.00962.x.

    ArticlePubMed Google Scholar

  4. Başkent, D. (2012). Effect of speech degradation on top-down repair: Phonemic restoration with simulations of cochlear implants and combined electric–acoustic stimulation. Journal of the Association for Research in Otolaryngology,13(5), 683–692. doi:10.1007/s10162-012-0334-3.

    ArticlePubMedPubMed Central Google Scholar

  5. Bloom, J. S., & Hynd, G. W. (2005). The role of the corpus callosum in interhemispheric transfer of information: excitation or inhibition? Neuropsychology Review,15(2), 59–71. doi:10.1007/s11065-005-6252-y.

    ArticlePubMed Google Scholar

  6. Bowman, H., & Wyble, B. (2007). The simultaneous type, serial token model of temporal attention and working memory. Psychological Review,114(1), 38–70. doi:10.1037/0033-295X.114.1.38.

    ArticlePubMed Google Scholar

  7. Brainard, D. H. (1997). The psychophysics toolbox. Spatial Vision,10, 433–436. doi:10.1163/156856897X00357.

    Article Google Scholar

  8. Chanmugam, A., Triplett, P., & Kelen, G. (2013). Emergency Psychiatry. Cambridge: Cambridge University Press.

    Book Google Scholar

  9. Chen, T., & Naveh-Benjamin, M. (2012). Assessing the associative deficit of older adults in long-term and short-term/working memory. Psychology and Aging,27(3), 666–682. doi:10.1037/a0026943.

    ArticlePubMed Google Scholar

  10. Coltheart, M. (1980). Iconic memory and visible persistence. Perception & Psychophysics,27, 183–228. doi:10.3758/BF03204258.

    Article Google Scholar

  11. Craik, F. I., & Salthouse, T. A. (2011). The handbook of aging and cognition. Hove: Psychology Press.

    Book Google Scholar

  12. Di Lollo, V. (1980). Temporal integration in visual memory. Journal of Experimental Psychology: General,109(1), 75–97. doi:10.1037/0096-3445.109.1.75.

    Article Google Scholar

  13. Di Lollo, V., Arnett, J. L., & Kruk, R. V. (1982). Age-related changes in rate of visual information processing. Journal of Experimental Psychology: Human Perception and Performance,8(2), 225–237. doi:10.1037/0096-1523.8.2.225.

    ArticlePubMed Google Scholar

  14. Di Lollo, V., Hogben, J. H., & Dixon, P. (1994). Temporal integration and segregation of brief visual-stimuli—Patterns of correlation in time. Perception & Psychophysics,55, 373–386.

    Article Google Scholar

  15. Fogerty, D., Humes, L. E., & Busey, T. A. (2016). Age-related declines in early sensory memory: Identification of rapid auditory and visual stimulus sequences. Frontiers in Aging Neuroscience. doi:10.3389/fnagi.2016.00090.

    ArticlePubMedPubMed Central Google Scholar

  16. Füllgrabe, C. (2013). Age-dependent changes in temporal-fine-structure processing in the absence of peripheral hearing loss. American Journal of Audiology,22(2), 313–315. doi:10.1044/1059-0889(2013/12-0070).

    ArticlePubMed Google Scholar

  17. Gehr, S. E., & Sommers, M. S. (1999). Age differences in backward masking. The Journal of the Acoustical Society of America,106(5), 2793–2799. doi:10.1121/1.428104.

    ArticlePubMed Google Scholar

  18. Gordon-Salant, S., & Fitzgibbons, P. J. (2001). Sources of age-related recognition difficulty for time-compressed speech. Journal of Speech, Language, and Hearing Research,44, 709–719. doi:10.1044/1092-4388(2001/056).

    ArticlePubMed Google Scholar

  19. Grady, C. (2012). The cognitive neuroscience of ageing. Nature Reviews Neuroscience,13(7), 491–505. doi:10.1038/nrn3256.

    ArticlePubMedPubMed Central Google Scholar

  20. Hogben, J. H., & di Lollo, V. (1974). Perceptual integration and perceptual segregation of brief visual stimuli. Vision Research,14(11), 1059–1069. doi:10.1016/0042-6989(74)90202-8.

    ArticlePubMed Google Scholar

  21. Horváth, J., Czigler, I., Winkler, I., & Teder-Sälejärvi, W. A. (2007). The temporal window of integration in elderly and young adults. Neurobiology of Aging,28(6), 964–975. doi:10.1016/j.neurobiolaging.2006.05.002.

    ArticlePubMed Google Scholar

  22. Humes, L. E., Busey, T. A., Craig, J. C., & Kewley-Port, D. (2009). The effects of age on sensory thresholds and temporal gap detection in hearing, vision, and touch. Attention, Perception, & Psychophysics,71(4), 860–871. doi:10.3758/APP.71.4.860.

    Article Google Scholar

  23. Humes, L. E., Busey, T. A., Craig, J., & Kewley-Port, D. (2013). Are age-related changes in cognitive function driven by age-related changes in sensory processing? Attention, Perception, & Psychophysics,75(3), 508–524. doi:10.3758/s13414-012-0406-9.

    Article Google Scholar

  24. International Council of Ophthalmology. (2002). Resources: Visual standards—aspects and ranges of vision loss. (n.d.). http://www.icoph.org/resources/10/Visual-Standards—Aspects-and-Ranges-of-Vision-Loss.html. Retrieved 7 February 2017.

  25. Klatt, D. H. (1980). Software for a cascade/parallel formant synthesizer. The Journal of the Acoustical Society of America,67(3), 971–995. doi:10.1121/1.383940.

    Article Google Scholar

  26. Klatt, D. H., & Klatt, L. C. (1990). Analysis, synthesis, and perception of voice quality variations among female and male talkers. The Journal of the Acoustical Society of America,87(2), 820–857. doi:10.1121/1.398894.

    ArticlePubMed Google Scholar

  27. Kleiner, M., Brainard, D., Pelli, D., Ingling, A., Murray, R., Broussard, C., et al. (2007). What’s new in Psychtoolbox-3. Perception,36, 14.

    Google Scholar

  28. Kline, D. W., Ikeda, D. M., & Schieber, F. J. (1982). Age and temporal resolution in color vision: When do red and green make yellow?. Journal of Gerontology,37(6), 705–709. doi:10.1093/geronj/37.6.705.

    ArticlePubMed Google Scholar

  29. Kline, D. W., & Orme-Rogers, C. (1978). Examination of stimulus persistence as the basis for superior visual identification performance among older adults. Journal of Gerontology,33(1), 76–81. doi:10.1093/geronj/33.1.76.

    ArticlePubMed Google Scholar

  30. Kolodziejczyk, I., & Szelag, E. (2008). Auditory perception of temporal order in centenarians in comparison with young and elderly subjects. Acta Neurobiologiae Experimentalis,68(3), 373–381.

    PubMed Google Scholar

  31. Kubovy, M. (1988). Should we resist the seductiveness of the space: time: vision: Audition analogy. Journal of Experimental Psychology: Human Perception and Performance,14(2), 318–320. doi:10.1037/0096-1523.14.2.318.

    Article Google Scholar

  32. Lahar, C. J., Isaak, M. I., & McArthur, A. D. (2001). Age differences in the magnitude of the attentional blink. Aging, Neuropsychology, and Cognition,8(2), 149–159. doi:10.1076/anec.

    Article Google Scholar

  33. Lindenberger, U., & Baltes, P. B. (1994). Sensory functioning and intelligence in old age: A strong connection. Psychology and Aging,9, 339–355. doi:10.1037//0882-7974.9.3.339.

    ArticlePubMed Google Scholar

  34. Lu, P. H., Lee, G. J., Raven, E. P., Tingus, K., Khoo, T., Thompson, P. M., & Bartzokis, G. (2011). Age-related slowing in cognitive processing speed is associated with myelin integrity in a very healthy elderly sample. Journal of Clinical and Experimental Neuropsychology,33(10), 1059–1068. doi:10.1080/13803395.2011.595397.

    ArticlePubMedPubMed Central Google Scholar

  35. Lu, P. H., Lee, G. J., Tishler, T. A., Meghpara, M., Thompson, P. M., & Bartzokis, G. (2013). Myelin breakdown mediates age-related slowing in cognitive processing speed in healthy elderly men. Brain and Cognition,81(1), 131–138. doi:10.1016/j.bandc.2012.09.006.

    ArticlePubMed Google Scholar

  36. Madden, D., & Allen, P. A. (2015). History of Cognitive Slowing Theory and Research. In N. A. Pachana (Ed.), Encyclopedia of Geropsychology (pp. 1–10). Singapore: Springer.

    Google Scholar

  37. Martini, A., & Mazzoli, M. (1999). Achievements of the European working group on genetics of hearing impairment. International Journal of Pediatric Otorhinolaryngology,49(S1), 155–158.

    Article Google Scholar

  38. Näätänen, R., Kujala, T., & Winkler, I. (2011). Auditory processing that leads to conscious perception: A unique window to central auditory processing opened by the mismatch negativity and related responses. Psychophysiology,48(1), 4–22. doi:10.1111/j.1469-8986.2010.01114.x.

    ArticlePubMed Google Scholar

  39. O’Callaghan, C. (2008). Object perception: Vision and audition. Philosophy Compass,3(4), 803–829.

    Article Google Scholar

  40. Pan, W. (2001). Akaike’s information criterion in generalized estimating equations. Biometrics,57, 120–125. doi:10.1111/j.0006-341X.2001.00120.x.

    ArticlePubMed Google Scholar

  41. Pelli, D. G. (1997). The VideoToolbox software for visual psychophysics: Transforming numbers into movies. Spatial Vision,10(4), 437–442.

    Article Google Scholar

  42. Poeppel, D. (2003). The analysis of speech in different temporal integration windows: cerebral lateralization as ‘asymmetric sampling in time’. Speech Communication,41, 245–255. doi:10.1016/S0167-6393(02)00107-3.

    Article Google Scholar

  43. Pols, L. C., Tromp, H. R., & Plomp, R. (1973). Frequency analysis of Dutch vowels from 50 male speakers. The Journal of the Acoustical Society of America,53(4), 1093–1101. doi:10.1121/1.1913429.

    ArticlePubMed Google Scholar

  44. Raymond, J. E., Shapiro, K. L., & Arnell, K. M. (1992). Temporary suppression of visual processing in an RSVP task: An attentional blink? Journal of Experimental Psychology: Human Perception and Performance,18(3), 849–860. doi:10.1037//0096-1523.18.3.849.

    ArticlePubMed Google Scholar

  45. Roberts, K. L., & Allen, H. A. (2016). Perception and cognition in the ageing brain: A brief review of the short- and long-term links between perceptual and cognitive decline. Frontiers in Aging Neuroscience. doi:10.3389/fnagi.2016.00039.

    ArticlePubMedPubMed Central Google Scholar

  46. Saija, J. D., Akyürek, E. G., Andringa, T. C., & Başkent, D. (2014a). Perceptual restoration of degraded speech is preserved with advancing age. Journal of the Association for Research in Otolaryngology,15, 139–148. doi:10.1037/a0035146.

    ArticlePubMed Google Scholar

  47. Saija, J. D., Andringa, T. C., Başkent, D., & Akyürek, E. G. (2014b). Temporal integration of consecutive tones into synthetic vowels demonstrates perceptual assembly in audition. Journal of Experimental Psychology: Human Perception and Performance,40(2), 857–869. doi:10.1037/a0035146.


Sours: https://link.springer.com/article/10.1007/s00426-017-0912-4
  1. Wooden crate chairs
  2. Futurama wiki
  3. U joint ratchet
  4. French imperialism timeline
  5. Avatar episode 41

Auditory temporal integration and the power function model

The auditory temporal integration function was studied with the objective of improving both its quantitative description and the specification of its principle independent variable, stimulus duration. In Sec. I, temporal integration data from 20 studies were subjected to uniform analyses using standardized definitions of duration and two models of temporal integration. Analyses revealed that these data were best described by a power function model used in conjunction with a definition of duration, termed assigned duration, that de-emphasized the rise/fall portions of the stimuli. There was a strong effect of stimulus frequency and, in general, the slope of the temporal integration function was less than 10 dB per decade of duration; i.e., a power function exponent less than 1.0. In Sec. II, an experimental study was performed to further evaluate the models and definitions. Detection thresholds were measured in 11 normal-hearing human subjects using a total of 24 single-burst and multiple-burst acoustic stimuli of 3.125 kHz. The issues addressed are: the quantitative description of the temporal integration function; the definition of stimulus duration; the similarity of the integration processes for single-burst and multiple-burst stimuli; and the contribution of rise/fall time to the integration process. A power function in conjunction with the assigned duration definition was again most effective in describing the data. Single- and multiple-burst stimuli both seemed to be integrated by the same central mechanism, with data for each type of stimulus being described by a power function exponent of approximately 0.6 at 3.125 kHz. It was concluded that the contribution of the rise/fall portions of the stimuli can be factored out from the rest of the temporal integration process. In Sec. III, the conclusions that emerged from the review of published work and the present experimental work suggested that auditory temporal integration is best described by a power function in conjunction with the assigned duration definition. The exponent for the power function is typically less than 1.0, and varies with frequency and hearing level. Second, a means of empirically assaying the contribution of the rise-fall portions of the stimuli is presented and evaluated. Finally, properties of a central auditory integrator are hypothesized.(ABSTRACT TRUNCATED AT 250 WORDS)

Sours: https://pubmed.ncbi.nlm.nih.gov/2212302/
Graded Potentials, EPSPs, IPSPs and Summation

Temporal Integration

Last Updated on Wed, 06 Jan 2021 | Hearing Loss

The term temporal integration (TI) refers to summation of stimulus intensity during the duration of the stimulus. As duration increases, a sensation like loudness increases, or the sound level at which the stimulus can be detected decreases. The stimuli may be various types of signals, such as tones or bands of noise. Similarly, short succeeding stimuli can combine their energies and provide a lower detection level than individual stimuli. The

A,: L, = 10*log(300/tA1+1)

A2: L, = 10*log(300/(t*0.8)+1)

Br- L, = -10*log(1 -exp(-t/300))

S2. Lt = -10*log(1-exp(-t/100)) _


I I ---L ILÜ —-

10° 101 102 T 103 104


10° 101 102 T 103 104


Figure 1. Temporal integration curves according to the functions shown in the legend. In curves A1 and A2, the time constant t = 300 ms; in A1, exponent m — 1; in A2, m — 0.8. In B1, t — 300 ms; in B2, t — 100 ms. The value of t — 300 ms is indicated by a mark on the abscissa.

TI has a time limit. For a stimulus longer than this limit, the loudness, or the detection (threshold) level, remains relatively constant.

Interest in studying TI is fueled by the need to understand auditory processing of speech—a signal that, by its nature, changes rapidly in time. Better understanding of the temporal characteristics of hearing should help us improve means for enhancement of speech communication in unfavorable listening environments, and of listeners with impaired hearing.

Graphs of the relationship between the stimulus duration (plotted on the horizontal coordinate, usually in milliseconds with a logarithmic scale) and the intensity level at the threshold of hearing (plotted on the vertical coordinate in decibels, dB) are called temporal integration curves (TICs). Examples of TICs are shown in Figure 1. The detection intensity level first declines as the stimulus duration increases and then, beyond a time limit called the critical duration, remains constant. The magnitude of TI can be expressed by the difference between the detection levels of long and short signals. The rate of decline of TICs is represented by slopes of the curves, which too are often used as indicators of TI magnitude. These slopes (they are negative) are usually expressed as the ratio of the change of level (in dB) per tenfold increase in signal duration [(L2-L1)/dec], or per doubling of signal duration. The slopes of the TICs and the values of the critical duration represent summary characteristics of TI. (The critical duration depends on a time constant t, a parameter in formulas describing TICs.)

Factors Affecting TI

The slope of the curves and the time constant depend on various factors, such as signal frequency, status of hear ing, and type of signals. The effect of signal frequency on TI is pronounced (Gerken, Bhat, and Hutchison-Clutter, 1990) and depends on signal duration. TI, as well as t, is greater at lower frequencies than at higher ones (e.g., Fastl, 1976; Nabelek, 1978; Florentine, Fastl, and Buus, 1988). At low frequencies and signal durations below 10 ms, the TIC slopes were found to be up to —15 dB/ dec (e.g., Green, Birdsall, and Tanner, 1957). At frequencies between 1 and 8 kHz and at signal durations between 20 and 100 ms, the slopes are between —10 and —8 dB/dec (e.g., Zwislocki, 1969; Gerken, Bhat, and Hutchison-Clutter, 1990). The steeper slopes at short signal durations as compared to slopes at longer signal durations are attributed to the loss of contribution of some energy due to spectral broadening, or ''splatter.'' When the frequency during the signal is not constant but is increasing, the slope between 20 ms and 80 ms of signal duration is smaller, about —9 dB/dec, than when the frequency is decreasing—about —13 dB/dec (Nabelek, 1978). For broadband masking conditions, the values of TI for constant tones are similar to those without masking, but some influence of the level of the masker was observed. This influence depends on signal duration. For signal durations between 2 and 10 ms, the TICs at medium masker levels are steeper than at low or high masker levels, and for signal durations over 20 ms, the TI values are not affected by the masker level (Oxenham, Moore, and Vickers, 1997). Formby et al. (1994) investigated the influence of bandwidth of a noise signal masked by an uncorrelated broadband noise on TI and t. They found that both TI and t were related inversely to the bandwidth, if the bandwidth was greater than the critical band of hearing (CB), and were relatively invariant if the bandwidth was smaller than the critical band. For gated signal and masker, Formby et al. (1994) identified at least three cues for signal detection: (1) a relative timing cue, (2) a spectral shape cue, and (3) a traditional energy cue. The timing and spectral shape cues count most for the shortest (10 ms) and narrowest (bandwidth = 63 Hz) signals, respectively. When the signal is a series of tone pulses, and not single bursts, the change of time interval between the pulses produces smaller change of TI than the change in duration of single bursts (Carlyon, Buus, and Florentine, 1990).

Listeners with hearing impairment generally show less temporal integration than listeners with normal hearing (e.g., Watson and Gengel, 1969; Gerken, Bhat, and Hutchison-Clutter, 1990). No effect of level of a broadband masker was found for listeners with impaired hearing at any signal duration.

Loudness increases when the duration of a short signal increases. When the signal level changes, the TI for loudness changes; however, the change is not monotonic (Buus, Florentine, and Poulsen, 1999). The change is greatest at moderate sensation levels and depends on signal duration. The effect of signal level on TI of loudness is greater at short than at long signal durations.

Donaldson, Viemeister, and Nelson (1997) found that TICs for electrical stimulation with the Nucleus-22 elec trode cochlear implant were considerably less steep than —8 dB/dec typically observed with acoustical stimulation. The slopes varied widely across subjects and across stimulated electrodes. When Shannon and Otto (1990) used a device called the auditory brainstem implant (ABI) and positioned its electrodes near the cochlear nucleus of listeners, they obtained only a shallow TIC over the range of 2- to 1000-ms signal duration.


A number of models for temporal integration have been proposed. The theoretical foundations for the mathematical description of TI are either deterministic or probabilistic. Deterministic models include power function models or exponential function models. One of the deterministic models is described mathematically by the power function t(It — Iy) = Iyt = const (Hughes, 1946), or in its more general form by Itm = C (Green et al., 1957). In these equations t is the stimulus duration, It is the threshold intensity at t, Iy is the threshold intensity for very long stimuli, t is the time constant of the integration process, m is the power function exponent, and C is a constant. The exponent m determines the slope of the curves (A1 and A2 in Fig. 1). The slope —3 dB/doubling or —10 dB/dec corresponds to m = 1. Another model is the exponential one It/Iy = 1/(1 — e—th) (Feldkeller and Oetinger, 1956; Plomp and Bouman, 1959). The curves B1 and B2 in Figure 1 correspond to this equation.

Zwislocki (1960) developed a temporal summation theory for two pulses separated in time and proposed a theory of TI for loudness (Zwislocki, 1969). In his model it is assumed that (1) a linear temporal integrator (with a time constant on the order of 200 ms) exists in the central nervous system; (2) a nonlinear transformation that produces compression precedes the temporal summation; and (3) neural excitation decreases exponentially with a short time constant at the input to the integrator that summates the central neural activity. (This last assumption indicates that the term temporal integration should not be interpreted as the integration of acoustic energy per se.)

Attempts to resolve an apparent discrepancy between high temporal resolution of hearing and long time constants of temporal integration have led to a number of models employing short integration times (e.g., Penner, 1978; Oxenham, Moore, and Vickers, 1997). Viemeister and Wakefield (1991) have not considered this discrepancy to be a real problem. Their model is based on a statistical probability approach and assumes multiple sampling. Taking their own data into account, Vie-meister and Wakefield conclude that power integration occurs only for pulses separated in time by less than about 5 ms, and that therefore their data are inconsistent with the classical view of TI involving long-term integration. However, they find the data to be compatible with the notion that the input is sampled at a fairly high rate and the obtained samples (or ''looks'') are stored in memory; while in the memory, the ''looks''

can be selectively accessed, weighted, and otherwise processed.

Dau, Kollmeier, and Kohlrausch (1997) proposed a multichannel model. They describe the effects of spectral and temporal integration in amplitude-modulation detection for a stochastic noise carrier. The model is based on the concept of a modulation filter-bank. To integrate information across frequency, the detection process in the model combines cues from all filters with an optimal decision statistic. To integrate information across time, a "multiple-look" strategy, similar to that proposed by Viemeister and Wakefield (1991), is realized within the detection stage of the model. The temporal integration involves a template that provides the basis for the optimal detector of the model. The length and the time constant of the template are variable: they change according to the task which the listener has to perform.

Although an extensive knowledge of temporal integration has been attained, many aspects of TI await further investigation. For example, evidence of some TI mechanism at a higher stage of the auditory pathway was found by Uppenkamp, Fobel, and Patterson (2001) when they compared the perception of short-frequency sweeps and the physiological response to them in the brainstem. The improved understanding of TI should provide a sounder basis for the development of means for securing better speech communication in general and for listeners with special problems, like those with coch-lear implants, in particular. Presently, TI studies are not limited to traditional topics but also cover higher levels of the brain, like the role of TI in establishing neural representations of phonemes (Tallal et al., 1998), and investigation of an association between a deficient TI and mental disturbances in schizophrenia (Haig et al., 2000; Michie, 2001).


Many thanks to Assoc. Prof. Lana S. Dixon for her help in securing pertinent references.

See also cLinicaL decision anaLysis; cochLear impLants; masking; temporal resolution.

—Igor V. Nabelek References

Buus, S., Florentine, M., and Poulsen, T. (1999). Temporal integration of loudness in listeners with hearing losses of primarily cochlear origin. Journal of the Acoustical Society of America, 105, 3464-3480. Carlyon, R. P., Buus, S., and Florentine, M. (1990). Temporal integration of trains of tone pulses by normal and by coch-learly impaired listeners. Journal of the Acoustical Society of America, 87, 260-268. Donaldson, G. S., Viemeister, N. F., and Nelson, D. A. (1997). Psychometric functions and temporal integration in electric hearing. Journal of the Acoustical Society of America, 101, 3706-3721.

Dau, T., Kollmeier, B., and Kohlrausch, A. (1997). Modeling auditory processing of amplitude modulation: II. Spectral and temporal integration. Journal of the Acoustical Society of America, 102, 2906-2919.

Fastl, H. (1976). Influence of test tone duration on auditory masking patterns. Audiology, 15, 63-71.

Feldkeller, R., and Oetinger, R. (1956). Die Horbarkeits-grenzen von Impulsen verschiedener Dauer. Acustica, 6, 481-493.

Florentine, M., Fastl, H., and Buus, S. (1988). Temporal integration in normal hearing, cochlear impairment, and impairment simulated by masking. Journal of the Acoustical Society ofAmerica, 84, 195-203.

Formby, C., Heinz, M. G., Luna, C. E., and Shaheen, M. K. (1994). Masked detection thresholds and temporal integration for noise band signals. Journal of the Acoustical Society of America, 96, 102-114.

Gerken, G. M., Bhat, V. K., and Hutchison-Clutter, M. (1990). Auditory temporal integration and the power function model. Journal of the Acoustical Society of America, 88, 767-778.

Green, D. M., Birdsall, T. G., and Tanner, W. P., Jr. (1957). Signal detection as a function of signal intensity and duration. Journal of the Acoustical Society of America, 29, 523531.

Haig, A. R., Gordon, E., De-Pascalis, V., Meares, R. A., Bahramali, H., and Harris, A. (2000). Gamma activity in schizophrenia: Evidence of impaired network binding? Clinical Neurophysiology, 111, 1461-1468.

Hughes, J. W. (1946). The threshold of audition for short periods of stimulation. Proceedings of the Royal Society of London. Series B: Biological Sciences, B133, 486-490.

Michie, P. T. (2001). What has MMN revealed about the auditory system in schizophrenia? International Journal of Psychophysiology, 42, 177-194.

Nabelek, I. V. (1978). Temporal summation of constant and gliding tones at masked auditory threshold. Journal of the Acoustical Society of America, 64, 751-763.

Oxenham, A. J., Moore, B. C., and Vickers, D. A. (1997). Short-term temporal integration: Evidence for the influence of peripheral compression. Journal of the Acoustical Society ofAmerica, 101, 3676-3687.

Penner, M. J. (1978). A power law transformation resulting in a class of short-term integrators that produce time-intensity trades for noise bursts. Journal of the Acoustical Society of America, 63, 195-201.

Plomp, R., and Bouman, M. A. (1959). Relation between hearing threshold and duration for tone pulses. Journal of the Acoustical Society of America, 31, 749-758.

Shannon, R. V., and Otto, S. R. (1990). Psychophysical measures from electrical stimulation of the human cochlear nucleus. Hearing Research, 47, 159-168.

Tallal, P., Merzenich, M. M., Miller, S., and Jenkins, W. (1998). Language learning impairments: Integrating basic science, technology, and remediation. Experimental Brain Research, 123, 210-219.

Uppenkamp, S., Fobel, S., and Patterson, R. D. (2001). The effects of temporal asymmetry on the detection and perception of short chirps. Hearing Research, 158, 71-83.

Viemeister, N. F., and Wakefield, G. H. (1991). Temporal integration and multiple looks. Journal of the Acoustical Society of America, 90, 858-865.

Watson, C. S., and Gengel, R. W. (1969). Signal duration and signal frequency in relation to auditory sensitivity. Journal of the Acoustical Society of America, 46, 989-997.

Zwislocki, J. J. (1960). Theory of temporal auditory summation. Journal of the Acoustical Society of America, 32, 1046-1060.

Zwislocki, J. J. (1969). Temporal summation of loudness: An analysis. Journal of the Acoustical Society of America, 46, 431-441.

Further Readings

Algom, D., Rubin, A., and Cohen-Raz, L. (1989). Binaural and temporal integration of the loudness of tones and noises. Perception and Psychophysics, 46, 155-166.

Bacon, S. P., Hicks, M. L., and Johnson, K. L. (2000). Temporal integration in the presence of off-frequency maskers. Journal of the Acoustical Society of America, 107, 922-932.

Buus, S. (1999). Temporal integration and multiple looks, revisited: Weights as a function of time. Journal of the Acoustical Society of America, 105, 2466-2475.

Cacace, A. T., Margolis, R. H., and Relkin, E. M. (1991). Threshold and suprathreshold temporal integration effects in the crossed and uncrossed human acoustic stapedius reflex. Journal of the Acoustical Society of America, 89, 12551261.

Csepe, V., Pantev, C., Hoke, M., Ross, B., and Hampson, S. (1997). Mismatch field to tone pairs: Neuromagnetic evidence for temporal integration at the sensory level. Electro-encephalography and Clinical Neurophysiology, 104, 1-9.

Fu, Q. J., and Shannon, R. V. (2000). Effect of stimulation rate on phoneme recognition by Nucleus-22 cochlear implant listeners. Journal of the Acoustical Society of America, 107, 589-597.

Garner, W. R., and Miller, G. A. (1947). The masked threshold of pure tones as a function of duration. Journal of Experimental Psychology, 37, 293.

Hall, J. W., and Fernandes, M. A. (1983). Temporal integration, frequency resolution, and off-frequency listening in normal-hearing and cochlear-impaired listeners. Journal of the Acoustical Society of America, 74, 1172-1177.

Kollmeier, B. (Ed.). (1995). Psychoacoustics, speech, and hearing aids. Proceedings of the summer school and international symposium, Bad Zwischenahn, Germany, 31 Aug.-5 Sept. 1995. World Scientific Singapore, 1996.

Moore, B. C. J. (1997). An introduction to the psychology of hearing (4th ed.). London: Academic Press.

Oxenham, A. J. (1998). Temporal integration at 6 kHz as a function of masker bandwidth. Journal of the Acoustical Society of America, 103, 1033-1042.

Sheft, S., and Yost, W. A. (1990). Temporal integration in amplitude modulation detection. Journal of the Acoustical Society ofAmerica, 88, 796-805.

Yost, W. A. (1991). Fundamentals of hearing: An introduction. San Diego, CA: Academic Press.

tion have focused on intensity variations in an attempt to separate purely temporal from spectro-temporal resolving capabilities (see auditory scene analysis). Temporal resolution is limited by auditory inertia resulting from mechanical and/or electrophysiological transduc-tion processes. Such a limitation effectively smoothes or attenuates the intensive changes of a stimulus, which reduces the salience of those changes. Impaired temporal resolution may be conceptualized as an increase in this smoothing process, and thus a loss of temporal information.

The influence of hearing impairment on temporal resolution depends on the site of lesion. For example, conductive hearing loss is often modeled as a simple attenuation characteristic and thus should not alter temporal resolution, given sufficient stimulus levels. Damage at the level of the cochlea, however, involves more than attenuation. Reduced outer hair cell function is associated with a reduction in sensitivity, frequency selectivity, and compression at the level of the basilar membrane. Each of these might influence the perception of intensity changes. For example, a loss of basilar membrane compression might provide a more salient representation of intensity changes and thus lead to improved performance on temporal resolution tasks involving such changes. Reduced frequency selectivity is analogous to broadening of a filter characteristic, which is associated with a shorter temporal response. This too might lead to improved temporal resolution. A loss of inner hair cell function, however, would reduce the quality and amount of information transmitted to the central auditory pathway, and might therefore lead to poor coding of temporal features. The altered neural function associated with a retrocochlear lesion may also lead to a less faithful representation of the temporal features of a sound.

Numerous techniques have been used to probe temporal resolution abilities; however, the two most common techniques are temporal gap detection and amplitude modulation detection (Fig. 1). Following the

Interval 1

Interval 2

Continue reading here: Temporal Resolution

Was this article helpful?

Sours: https://www.flandershealth.us/hearing-loss-2/temporal-integration.html

Definition temporal integration

Itself began to weed out its outfits. During the selection process, I tried on clothes. The sight of a naked beloved woman aroused a desire that overpowered the guy's minor mood. Darling, let me see how your gates healed. making some effort of the resting sister, the guy pulled her to him, kissed her on the lips.

Temporal structure in perception and attention - Dr. Ayelet N. Landau

Exclaimed Marina when we flew in through the opening in the wall she opened. And they saw the astronauts frozen in their chairs like dolls. There were about a hundred of them, and they all stared ahead at the huge screen. On which was seen a planet streaked with craters from the fall of meteorites on it.

And what surprised me was that the crew members of the ship, sitting motionless in their chairs in front of the observation window.

You will also like:

The city. Sorry if there are any mistakes. It is not possible to notice everything ) P.

7668 7669 7670 7671 7672