Rapid calibration to dynamic temporal contexts

The prediction of future events and the preparation of appropriate behavioural reactions rely on an accurate perception of temporal regularities. In dynamic environments, temporal regularities are subject to slow and sudden changes, and adaptation to these changes is an important requirement for efficient behaviour. Bayesian models have proven a useful tool to understand the processing of temporal regularities in humans; yet an open question pertains to the degree of flexibility of the prior that is required for optimal modelling of behaviour. Here we directly compare dynamic models (with continuously changing prior expectations) and static models (a stable prior for each experimental session) with their ability to describe regression effects in interval timing. Our results show that dynamic Bayesian models are superior when describing the responses to slow, continuous environmental changes, whereas static models are more suitable to describe responses to sudden changes. In time perception research, these results will be informative for the choice of adequate computational models and enhance our understanding of the neuronal computations underlying human timing behaviour.

Adaptive oscillators support Bayesian prediction in temporal processing

Humans excel at predictively synchronizing their behavior with external rhythms, as in dance or music performance. The neural processes underlying rhythmic inferences are debated: whether predictive perception relies on high-level generative models or whether it can readily be implemented locally by hard-coded intrinsic oscillators synchronizing to rhythmic input remains unclear and different underlying computational mechanisms have been proposed. Here we explore human perception for tone sequences with some temporal regularity at varying rates, but with considerable variability. Next, using a dynamical systems perspective, we successfully model the participants behavior using an adaptive frequency oscillator which adjusts its spontaneous frequency based on the rate of stimuli. This model better reflects human behavior than a canonical nonlinear oscillator and a predictive ramping model-both widely used for temporal estimation and prediction-and demonstrate that the classical distinction between absolute and relative computational mechanisms can be unified under this framework. In addition, we show that neural oscillators may constitute hard-coded physiological priors-in a Bayesian sense-that reduce temporal uncertainty and facilitate the predictive processing of noisy rhythms. Together, the results show that adaptive oscillators provide an elegant and biologically plausible means to subserve rhythmic inference, reconciling previously incompatible frameworks for temporal inferential processes.

Phasic and sustained interactions of multisensory interplay and temporal expectation

Every moment organisms are confronted with complex streams of information which they use to generate a reliable mental model of the world. There is converging evidence for several optimization mechanisms instrumental in integrating (or segregating) incoming information; among them are multisensory interplay (MSI) and temporal expectation (TE). Both mechanisms can account for enhanced perceptual sensitivity and are well studied in isolation; how these two mechanisms interact is currently less well-known. Here, we tested in a series of four psychophysical experiments for TE effects in uni- and multisensory contexts with different levels of modality-related and spatial uncertainty. We found that TE enhanced perceptual sensitivity for the multisensory relative to the best unisensory condition (i.e. multisensory facilitation according to the max-criterion). In the latter TE effects even vanished if stimulus-related spatial uncertainty was increased. Accordingly, computational modelling indicated that TE, modality-related and spatial uncertainty predict multisensory facilitation. Finally, the analysis of stimulus history revealed that matching expectation at trial n-1 selectively improves multisensory performance irrespective of stimulus-related uncertainty. Together, our results indicate that benefits of multisensory stimulation are enhanced by TE especially in noisy environments, which allows for more robust information extraction to boost performance on both short and sustained time ranges.

Musical Scales in Tone Sequences Improve Temporal Accuracy

Predicting the time of stimulus onset is a key component in perception. Previous investigations of perceived timing have focused on the effect of stimulus properties such as rhythm and temporal irregularity, but the influence of non-temporal properties and their role in predicting stimulus timing has not been exhaustively considered. The present study aims to understand how a non-temporal pattern in a sequence of regularly timed stimuli could improve or bias the detection of temporal deviations. We presented interspersed sequences of 3, 4, 5, and 6 auditory tones where only the timing of the last stimulus could slightly deviate from isochrony. Participants reported whether the last tone was ‘earlier’ or ‘later’ relative to the expected regular timing. In two conditions, the tones composing the sequence were either organized into musical scales or they were random tones. In one experiment, all sequences ended with the same tone; in the other experiment, each sequence ended with a different tone. Results indicate higher discriminability of anisochrony with musical scales and with longer sequences, irrespective of the knowledge of the final tone. Such an outcome suggests that the predictability of non-temporal properties, as enabled by the musical scale pattern, can be a factor in determining the sensitivity of time judgments.

Multiple Looks of Auditory Empty Durations Both Improve and Impair Temporal Sensitivity

Discrimination of two neighboring empty durations that are marked by three successive sounds is improved when the presentation of the first (standard, S) duration is repeated before that of the second (comparison, C), as SSSSC. This improvement in sensitivity, called the multiple-look effect, has been explained by a statistical model regarding variability. This model assumes that the perceived duration of the standard is averaged across observations (within a trial within an individual). The increasing of the number of observations thus reduces the standard error of the mean perceived duration. Alternatively, the multiple-look effect is attributed to the listener’s prediction based on regular rhythm. Listeners perceive regular rhythm during the repetition of the standard, predict the timing of subsequent sounds, and detect a sound that is displaced from the predicted timing. These models were tested in the present experiment in which the main factor was a temporal separation between the standard and the comparison; i.e., these durations were adjacent to each other as SSSSC or separated by a temporal blank as SSSS_C. The results differed between stimulus structures. First, the multiple-look effect was replicated in the SSSSC condition (yielding a higher performance than SC), but disappeared in SSSS_C (having no difference with S_C). Second, no multiple-look effect occurred in CSSSS (no difference with CS), and moreover, an impairment effect was observed in C_SSSS (a lower performance than C_S). Finally, discrimination was improved in SSSS_CCCC compared with SSSSCCCC, the effect being kept even when sounds were aligned at irregular intervals. These findings are not consistent with those expected from the statistical model because the temporal separation should have produced no effects if the number of standards had been a sole parameter determining the multiple-look effect. The prediction-based model can explain the first finding; inserting a blank between the standard and the comparison violates the listener’s prediction based on regular rhythm, thus reducing the multiple-look effect. However, it did not expect the other findings and required revisions. Notably, the second finding indicates that the formation of regular rhythm can impair temporal discrimination. In other words, an inversed multiple-look effect occurs.

Optimal Perceived Timing: Integrating Sensory Information with Dynamically Updated Expectations

The environment has a temporal structure, and knowing when a stimulus will appear translates into increased perceptual performance. Here we investigated how the human brain exploits temporal regularity in stimulus sequences for perception. We find that the timing of stimuli that occasionally deviate from a regularly paced sequence is perceptually distorted. Stimuli presented earlier than expected are perceptually delayed, whereas stimuli presented on time and later than expected are perceptually accelerated. This result suggests that the brain regularizes slightly deviant stimuli with an asymmetry that leads to the perceptual acceleration of expected stimuli. We present a Bayesian model for the combination of dynamically-updated expectations, in the form of a priori probability of encountering future stimuli, with incoming sensory information. The asymmetries in the results are accounted for by the asymmetries in the distributions involved in the computational process.