RESEARCH DOSSIER
Time Perception and Temporal Integration
Abstract. Time perception and temporal integration describe how organisms discriminate durations, infer the sequence of events, predict when information will arrive, and combine signals whose physical and neural delays differ. This research monograph offers a critical narrative synthesis of historical timing experiments, scalar expectancy and attentional accounts, Bayesian and neural-state models, prospective and retrospective judgments, temporal order and simultaneity procedures, sensory recalibration, multisensory binding, motor synchronization, and recent evidence on measurement and individual variation. The analysis distinguishes perceived duration from order, synchrony, passage, and sensorimotor timing; evaluates whether experimental tasks reveal common mechanisms; and examines the evidential limits of neural, behavioral, developmental, and computational claims. The overall synthesis supports interacting, task-dependent temporal computations, calibrated by uncertainty, attention, sensory modality, prior experience, and action. It records unresolved questions about latent structure, causal neural implementation, temporal binding metrics, transfer, and population generalization. This is a critical narrative synthesis of published research, not an original experiment or independently conducted systematic review.
Keywords: time perception; temporal integration; duration estimation; interval timing; scalar expectancy theory; psychophysical timing; prospective timing; retrospective timing; temporal order judgment; simultaneity judgment; temporal binding window; multisensory integration; temporal recalibration; Bayesian timing; neural population dynamics; temporal attention; sensorimotor synchronization; subjective duration; temporal prediction; Artificial Era
Introduction: Time as an Object of Psychological Measurement
Time perception and temporal integration concern the organization of experience and behavior with respect to duration, succession, simultaneity, rhythm, and the relations among events. An observer may discriminate two intervals accurately while misjudging the order of crossmodal signals, may synchronize movement to a periodic cue without verbal duration estimation, and may judge a remembered episode as long despite reporting that it passed quickly. These outcomes must be analyzed as different dependent variables rather than collapsed into one subjective clock. The central questions are how temporal information is encoded, retained, combined with prior experience, and used to guide perception and action. Duration judgment requires a relation between onset and offset; temporal order requires comparing event latencies; temporal integration requires a rule for binding information received across an interval. Behavioral psychophysics supplies methods for separating these operations Grondin (2010) Ivry & Schlerf (2008).
This monograph evaluates historical accounts of interval timing, scalar and attentional models, probabilistic and neural-state computations, multisensory simultaneity and order, temporal attention, sensorimotor prediction, and cross-task evidence. Its methodological approach is a critical narrative synthesis of published original experiments and major reviews. For each mechanism the analysis asks which manipulation discriminates the account from its competitors, whether model parameters are identifiable from the data, and how laboratory measurements generalize across modalities, contexts, development, and individual experience. No original participant data, preregistered systematic search, or meta-analysis is claimed. The emphasis is on convergent empirical constraints and substantive disagreements rather than a preferred single mechanism Merchant et al. (2013) Paton & Buonomano (2018).
Several broad principles organize the investigation. Physical intervals are measurable with calibrated instruments, whereas perceived time is inferred from judgments or behavior; neither behavioral response nor neural signal is a direct record of a unitary temporal experience. Temporal tasks operate at different scales, and the precision of a millisecond-order discrimination cannot be compared straightforwardly with memory for an entire day. Neural transmission delays, attention, decision criteria, and environmental statistics all influence temporal measurements. A defensible theory must explain how an organism stays behaviorally coordinated amid these differences, and must also state the circumstances under which an apparent timing effect is better explained by changes in non-temporal processing Coull et al. (2011) Eagleman et al. (2005).
Historical Development of Interval Timing and Psychological Chronometry
Scientific accounts of temporal judgment emerged through the nineteenth-century expansion of psychophysics, experimental chronology, and physiological measurement. The enduring problem was that equal physical intervals could be judged differently depending on their duration and experimental context. Twentieth-century chronometric work transformed this descriptive observation into explicit quantitative mechanisms. In a foundational formal account, Treisman (1963) described temporal discrimination using an internal-clock model with a pacemaker, a counter-like stage, and psychological parameters governing variability. The internal-clock metaphor became influential because it generated testable predictions about the relation between interval length and discrimination, while the metaphor itself was not evidence that a literal neural counter had been discovered Treisman (1963) Grondin (2010).
Animal timing research provided the next major formal step. Gibbon (1977) established a scalar expectancy framework that connected temporal discrimination with a Weber-like relation between variability and duration. Gibbon, Church, and Meck (1984) developed a theory in which remembered temporal samples, comparison operations, and scalar variability jointly explained performance. Church (1984) reviewed properties required of internal-clock mechanisms. These theories became especially useful for interval bisection, peak procedures, and generalization tasks in which whole response distributions, rather than single thresholds, could be compared with mathematical predictions. The historical innovation was the operational integration of behavior, memory, and decision into a model that could be rejected by data Gibbon (1977) Gibbon et al. (1984) Church (1984).
Subsequent work revealed how much of temporal performance depends on the timing task. Human experiments found interference between temporal estimation and concurrent tasks, differences between prospective and retrospective duration judgments, and context effects on reproduced intervals. In parallel, systems neuroscience examined cortico-striatal and cerebellar contributions, neural dynamics, and the role of perception and action. The history is therefore best understood as an accumulation of complementary and competing computational questions: how temporal precision scales, how events are compared in memory, how temporal evidence is updated, and why temporal judgments adapt to experimental distributions. Historical importance and current mechanistic adequacy are assessed separately throughout this study Brown (1997) Block & Zakay (1997) Buhusi & Meck (2005).
Defining Duration, Succession, Simultaneity, and Temporal Binding
An interval is the temporal separation between designated events or boundaries. Duration estimation assigns a magnitude to that interval, using verbal estimates, reproduction, production, comparison, or categorization. Temporal order judgment identifies which of two events came first; simultaneity judgment reports whether events are perceived as occurring together. Temporal binding denotes the integration of temporally distributed evidence into a coherent event or the compression of perceived separation between related action and consequence. Temporal attention selects moments at which information is expected, and sensorimotor synchronization coordinates action phase with an external sequence. The definitions identify overlapping tasks but different observables and computational demands Vroomen & Keetels (2010) Repp & Su (2013).
The phrase temporal integration is used at several levels of analysis. At the sensory level, information arriving during a finite period can be pooled into one perceptual estimate. At the multisensory level, signals from different modalities can be assigned to a common event despite unequal sensory delays. At the cognitive level, sequences may be grouped into episodes and used to predict future input. At the action level, proprioceptive and external consequences can be coupled in a sensorimotor representation. A temporal integration window is an experimentally estimated range over which a particular task yields increased combination, not necessarily a fixed interval of consciousness or an anatomical gate. Different tasks can yield different windows in the same observer Stevenson & Wallace (2013) Akyürek (2025).
Temporal precision, accuracy, bias, and stability are also distinct. A narrow distribution of judgments can be consistently offset from physical time; a nearly unbiased group mean can conceal considerable individual imprecision. The point of subjective simultaneity measures the offset at which two signals are most often judged synchronous, while a just-noticeable difference or fitted psychometric width measures discrimination. Estimates depend on modality, spatial configuration, stimulus complexity, trial history, instruction, and decision rule. A theory that explains a mean shift without explaining response distributions and criterion changes remains incomplete, particularly when translating between simultaneity and temporal-order tasks García-Pérez & Alcalá-Quintana (2012) O'Donohue et al. (2025).
Temporal Psychophysics: Experimental Control and Response Measures
Duration comparison presents standard and comparison intervals and asks observers which was longer. Temporal bisection presents learned short and long anchors and categorizes intermediate durations; generalization measures responses around an established reference; reproduction asks for an interval-matched action; production requests a specified subjective duration. Each procedure combines sensory encoding with working memory, decision, and motor execution in different proportions. For example, temporal reproduction adds movement initiation and termination variability to whatever duration representation was acquired. Bisection is sensitive to reference learning and response criterion, and a comparison task may reveal order-dependent biases. Cross-method agreement is therefore empirical evidence to be tested rather than assumed from the label timing Wearden (1991) Grondin (2010).
Temporal-order and simultaneity experiments often vary stimulus onset asynchrony over positive and negative offsets, then fit psychometric response functions. The investigator should report onset conventions, sensory delivery uncertainty, trial randomization, catch trials, fitted lapse and guess parameters, and whether the task demanded a speeded judgment. Apparatus timing can be critical: screen refresh, audio buffering, presentation software, headphones, and physical path delays change the stimulus the participant actually receives. Even with perfect delivery, a judgment of which event was first depends on sensory latency and response mapping. Model comparisons should retain trial-level responses and inferential uncertainty García-Pérez & Alcalá-Quintana (2012) Vroomen & Keetels (2010).
Reliability is especially important because an observed difference between groups may represent changes in threshold precision, response bias, or strategy. Within-observer adaptive procedures can concentrate observations around informative asynchronies, but the chosen algorithm and prior assumptions affect parameter recovery. Staircases and psychometric fitting require checks for reversals, nonstationarity, perceptual learning, and aftereffects. In tasks probing temporal binding, the observer can change criteria across blocks depending on whether instructions emphasize physical simultaneity or event identity. Reporting raw distributions alongside model-based indices improves interpretability and preserves the opportunity for independent reanalysis Stevenson & Wallace (2013) O'Donohue et al. (2025).
Temporal Scales and the Limits of a Single Timing Construct
Temporal processing extends from very short sensory integration periods to interval estimates lasting seconds or minutes, as well as anticipation across extended episodes. Short event-order judgments, duration bisection, and temporal prediction engage partly overlapping resources but need not use the same representational code. Ivry and Schlerf (2008) framed a central debate between dedicated timing machinery and intrinsic timing emerging from the evolving states of systems performing other computations. A model valid for interval reproduction cannot automatically explain a rapid audiovisual order judgment; an entrained neural oscillation does not automatically specify the magnitude of a remembered interval Ivry & Schlerf (2008) Mauk & Buonomano (2004).
An instructive distinction concerns rhythmic versus isolated-event timing. A periodic stimulus supports phase estimation and temporal expectation from a succession of regular cues. A single onset followed by an uncertain offset may instead require elapsed-time inference without recurrent correction. In sensorimotor performance, a participant can predict a beat, compare expected with actual auditory feedback, and correct movement phase. The timing of a single response interval may involve different variability and error dynamics. Repp and Su (2013) reviewed this rich literature and made clear that synchronization should be characterized by asynchronies, variability, and correction patterns rather than only average tapping rates Repp & Su (2013) Nobre & van Ede (2018).
Subjective passage is a further construct. Reports that time seemed to 'fly' or 'drag' during an episode may depend on attentional absorption, changes in stimulus density, emotion, and the distinction between momentary experience and subsequent memory. Such reports have different error structures from a two-alternative interval discrimination task. Block and Zakay's review of prospective and retrospective judgments provides a starting point for separating online monitoring from remembered duration. This distinction matters for ordinary language and for research on events in daily life, where different time reports can appear contradictory while referring to different psychological operations Block & Zakay (1997) Zakay & Block (2004).
Scientific Evidence and the Logic of Model Discrimination
The most informative timing experiment distinguishes predictions before interpreting a statistically significant contrast. A pacemaker account, an attentional-gate account, and a Bayesian context account may all produce longer duration reports after a stimulus manipulation. They can diverge in predicted effects on variance, adaptation, serial dependence, transfer to a second modality, or performance under a concurrent task. Experiments should therefore vary relevant dimensions independently and assess entire judgment distributions. Researchers should report the procedures used to estimate parameters, the uncertainty in recovered values, and whether alternative models achieve comparable fit with fewer assumptions Gibbon (1977) Jazayeri & Shadlen (2010).
Correlational neural recordings can identify timing-related populations or networks but cannot, by themselves, establish that a signal implements the psychological code. A neural signal might reflect attention, working memory, decision commitment, motor preparation, or elapsed time. Causal interventions, representational analyses, cross-condition generalization, and computational predictions offer stronger constraints, while each carries its own limitations. Similarly, a significant relationship between two timing tasks does not guarantee a common latent timing faculty when both tasks share response, vigilance, or memory demands. Measurement models must be explicit before theoretical constructs are inferred Merchant et al. (2013) Paton & Buonomano (2018).
The synthesis that follows evaluates evidence at multiple levels: formal discrimination laws, behavioral signatures of expectation and learning, sensory timing across modalities, neural network dynamics, and the organization of coordinated action. It favors a layered explanation in which temporal representations are shaped by task structure and statistical context, while remaining open to shared computational principles. This is an evidential position, not the introduction of a proprietary timing theory. The key test is whether a proposed mechanism explains additional phenomena with independently corroborated predictions and survives comparison against plausible non-temporal explanations Grondin (2010) Coull et al. (2011).
Scalar Timing, Weber Variability, and Internal-Clock Architectures
Scalar timing theory describes a statistical property of interval judgment: within relevant ranges, variability can grow approximately in proportion to the estimated duration. Gibbon's 1977 analysis expressed this Weber-like relation through scalar expectancy theory, and later work by Gibbon, Church, and Meck specified clock, memory, and decision contributions. At the level of the model, a pulse-generating process provides accumulating information, memory preserves variable references, and a comparison rule governs action. The separation is useful because a manipulation can affect pacemaker rate, reference storage, or decision threshold and produce different patterns of bias and spread. Scalar properties constitute a falsifiable claim about response distributions and their transformations Gibbon (1977) Gibbon et al. (1984).
A central empirical issue is how widely the scalar signature generalizes. Wearden (1991) investigated scalar properties in human timing, while Church (1984) addressed properties expected from clock-based models. Approximate scalar variability in a particular procedure supports constraints on models, although it does not uniquely identify pulses, counters, or any particular brain structure. In some tasks response variance contains non-temporal components, and different task ranges yield different slopes or departures from proportionality. A methodological assessment therefore plots variance against interval magnitude and compares models with additive sensory or motor noise, scalar memory noise, and changing decision criteria Wearden (1991) Church (1984).
The internal-clock family remains significant because it maps complex behavior to manipulable latent stages. Its challenges involve biological plausibility, interactions among timescales, and the number of separate mechanisms needed for modalities and tasks. A single pacemaker metaphor is insufficient when stimulus expectations alter the reproduction of identical intervals or when distributed networks learn temporal contingencies without explicit pulse counting. The contemporary question is which predictions from scalar models survive tests involving contextual distributions, neural population dynamics, attentional load, and cross-task transfer. These comparisons preserve the strongest mathematical contributions of scalar timing while avoiding unwarranted identification of a clock with observed firing activity Buhusi & Meck (2005) Paton & Buonomano (2018).
Attention, Cognitive Load, and the Allocation of Temporal Resources
Attention changes timing performance because temporal tasks compete with other processing demands. In a prospective duration judgment, observers know that time will be estimated, making temporal monitoring an available strategy. When they must simultaneously maintain information, discriminate stimuli, or make decisions, performance may shift as resources are redistributed. Brown (1997) experimentally examined interference between timing and concurrent working-memory demands, demonstrating that temporal estimation participates in a broader architecture of attention and executive resources. Such interference motivates attentional-gate models in which the amount of accumulated temporal evidence depends partly on how attention is allocated Brown (1997) Zakay & Block (2004).
Attention can change estimates in opposing directions according to task demands. A demanding distractor may reduce deliberate monitoring of a measured interval, whereas a salient or unexpected event can be experienced as lasting longer than a comparison. Tse and colleagues (2004) investigated the subjective expansion of time associated with attention-grabbing events. These effects are important because one might otherwise equate more attentional processing with a uniform increase in reported duration. A precise account asks what aspect of attention is manipulated—temporal monitoring, exogenous capture, feature processing, expectancy, or memory formation—and whether the dependent variable is online reproduction or retrospective report Tse et al. (2004) Kanai et al. (2006).
Attentional models must be tested against encoding and decision alternatives. A secondary task might delay perceived onset, alter memory for offset, change the criterion for responding, or increase motor initiation latency. Attentional-gate theories become more compelling when they predict changes in interval-response variance or when similar manipulations have separable consequences under prospective and retrospective instructions. Factorial designs that vary task difficulty, stimulus salience, and temporal relevance can isolate contributions more effectively than a single dual-task contrast. The broader implication is that timing recruits domain-general control in task-specific ways, with experimental details determining the observable direction of bias Brown (1997) Block & Zakay (1997).
Prospective and Retrospective Duration as Different Inference Problems
Prospective duration judgments arise when participants expect to estimate the interval before or during its occurrence. Retrospective judgments are obtained after an interval when observers did not anticipate the temporal question. Block and Zakay's (1997) meta-analytic review distinguished systematic differences between these conditions, and their later executive-control account emphasized the changing contributions of attention and memory. The distinction matters because apparent contradictions—an engaging interval seeming brief as it unfolds but rich and lengthy in recollection—can originate in different sources of evidence available at judgment time Block & Zakay (1997) Zakay & Block (2004).
During prospective timing, observers can monitor elapsed time and maintain a representation of the reference duration. Attention devoted to non-temporal material can interfere with this monitoring. During retrospective timing, reconstruction can draw on contextual changes, event boundaries, the density of remembered information, and an episode's organization. There is consequently no requirement that interventions increasing the number of memorable events also increase online duration estimates. This distinction helps formulate experiments in which the same experience is later queried in distinct ways and clarifies why everyday statements about fast and slow time need operational definition Block & Zakay (1997) Grondin (2010).
Retrospective reports are nevertheless empirical measurements, not inherently subjective anecdotes outside scientific analysis. Their validity improves when exposure conditions, event segmentation, delay to report, and memory probes are controlled, and when the judgment is compared against identifiable predictions about stored contextual information. The measurement problem is that autobiographical recollection, episodic detail, and duration inference may covary without being causally identical. Combining immediate and delayed measures on equivalent stimuli can reveal which effects survive memory consolidation or reinterpretation. No universal functional inversion between prospective and retrospective time should be assumed for every task, interval range, or population Zakay & Block (2004) Merchant et al. (2013).
Bayesian Calibration, Context, and Central-Tendency Effects
Temporal estimation is often conditioned by a distribution of possible intervals. When intervals in an experiment are drawn from a restricted range, reproduced durations may shift toward the center of that distribution, yielding overestimation of shorter intervals and underestimation of longer ones. Jazayeri and Shadlen (2010) showed that temporal context calibrates interval timing and analyzed behavior through a Bayesian framework. An observer with uncertain evidence can improve expected accuracy under a particular loss function by combining a noisy current measurement with a prior distribution learned from experience. The resulting biases are not merely failures of a clock; they can express structured statistical inference Jazayeri & Shadlen (2010).
A Bayesian observer account requires explicit assumptions about the prior, likelihood, cost of errors, and mapping from estimates to responses. Identical-looking central-tendency effects can also arise from response regression, memory constraints, or motor-related biases. The strength of Bayesian explanation lies in predicting how estimates change when the range, distribution, frequency, feedback, or uncertainty changes. Experimental contrasts should independently manipulate stimulus statistics and sensory precision and then compare posterior-based predictions against alternatives. The calibration of one temporal context does not establish that all timing tasks share an identical prior or computational implementation Jazayeri & Shadlen (2010) Paton & Buonomano (2018).
Uncertainty is a productive bridge between duration and temporal integration. The time at which a signal is inferred to have occurred may depend on known transmission delays, prior event probabilities, and the precision of each modality. The observer can calibrate audiovisual synchrony to a stable offset, and can change interval reproduction after learning new contextual statistics. These phenomena suggest that temporal judgments often involve a measurement followed by inference. Whether a common Bayesian computation explains duration reproduction and crossmodal binding is a testable cross-task hypothesis, whose evaluation must preserve the differences between interval magnitude, causal association, and event order Fujisaki et al. (2004) Shams & Beierholm (2022).
Neural-State and Distributed Models of Elapsed Time
Models of intrinsic timing propose that elapsed duration can be decoded from the evolving states of neural systems that are processing stimuli, maintaining information, or preparing actions. Karmarkar and Buonomano (2007) developed a network-state account in which the state of a recurrent neural circuit provided temporal information without a dedicated ticking counter. Mauk and Buonomano (2004) surveyed candidate neural mechanisms across temporal scales and argued for attention to the computations actually required by particular behaviors. In these approaches, recurrent dynamics, synaptic change, adaptation, and circuit trajectories can encode the passage of time indirectly through state evolution Karmarkar & Buonomano (2007) Mauk & Buonomano (2004).
Neural-state models generate different experimental expectations from a fully centralized clock. If the temporal code emerges from local stimulus-responsive dynamics, changes in sensory context or network history may alter interval estimates even when nominal duration is unchanged. Such accounts can explain why temporal precision differs between stimulus features and modalities, and why trained circuits sometimes exhibit specialized timescales. They must also explain how temporal representations remain robust across changes in input, how an observer transfers learning between contexts, and how the brain compares times encoded in distributed formats. Model simulations require empirical validation against neural and behavioral trajectories, rather than merely reproducing a selected psychometric curve Ivry & Schlerf (2008) Paton & Buonomano (2018).
The contrast between dedicated and intrinsic mechanisms is an organizing scientific question, not a requirement that all temporal functions belong to one side. A biological system may employ distributed sensory trajectories for one task and more specialized recurrent or cortico-striatal computations for another. Neural firing patterns that evolve across a trial provide candidate temporal codes, yet identifying their causal role requires perturbation and generalization tests. The strongest mechanistic programs explain precise behavioral outcomes through neural population changes and establish which variables the population represents under relevant controls. Timing theory thus gains from studying the operations required by each task before assigning anatomy Coull et al. (2011) Merchant et al. (2013).
Duration Distortions, Predictability, and Perceptual Event Structure
Subjective duration is sensitive to stimulus content, salience, movement, and predictability. Pariyadath and Eagleman (2007) tested effects of predictability on experienced duration, while Kanai and colleagues (2006) examined changes in perceived time in dynamic visual displays. Such findings show that equal physical durations can yield systematic differences in judgments, but the underlying causes may include adaptation, event expectation, temporal segmentation, and shifts in comparison standards. Each perceptual context should be specified rather than reduced to a general declaration that the brain slows or speeds time Pariyadath & Eagleman (2007) Kanai et al. (2006).
An oddball or unexpected stimulus in an otherwise predictable sequence may be judged as longer. Attention, increased neural response to novelty, reduced adaptation, and reconstruction of event boundaries are candidates for explanation. Experiments should determine whether the effect occurs in discrimination as well as explicit estimation, whether it transfers to other visual or auditory features, and whether it depends on surprise rather than physical salience. A salience manipulation can also shift onset latency; an apparent duration expansion could partly originate from the timing of perceptual markers. The issue is best addressed with converging task measures and explicit alternative models Tse et al. (2004) Pariyadath & Eagleman (2007).
Context effects also connect timing to the perceptual organization of events. In a dynamic scene, changes in motion, continuity, and grouping define what counts as the same event and which onset-offset pair is judged. Temporal integration can influence which stimulus boundaries are registered, while the identified boundaries in turn determine duration estimates. This mutual dependence prevents a simple interpretation of time perception as a passive readout from an independent clock. A complete account requires perceptual event parsing, contextual learning, and decision strategy in addition to elapsed-time representation Akyürek (2025) Grondin (2010).
Temporal Order and Simultaneity: Two Psychophysical Judgments
Temporal-order judgment and simultaneity judgment interrogate related but nonidentical operations. A temporal-order task asks whether sound preceded flash or flash preceded sound, often yielding an S-shaped psychometric function over stimulus onset asynchrony. A simultaneity task asks whether two stimuli occurred together, often yielding a peaked or asymmetric function. Both can reveal sensory and decisional biases, but their thresholds and points of subjective simultaneity need not coincide. García-Pérez and Alcalá-Quintana (2012) constructed a quantitative account of discrepancies across the tasks, showing why inferring a common temporal window directly from fitted widths is problematic García-Pérez & Alcalá-Quintana (2012) Vroomen & Keetels (2010).
Conceptually, a sensory signal arrives after variable processing delays whose distributions differ between sight, hearing, and touch. Temporal order can be inferred from a difference between noisy arrival-time estimates, while simultaneity may be judged using a different criterion or representation. A participant who often reports synchrony for a particular offset need not fail to discriminate the order of those same stimuli under explicit instruction. Conversely, response certainty in an order task can change without a proportional change in the subjective synchrony interval. Experimental designs that fit both response functions and model choice and lapse behavior provide stronger evidence than reporting a single unqualified temporal binding window García-Pérez & Alcalá-Quintana (2012) Stevenson & Wallace (2013).
Interpreting such measurements requires operational detail. Spatial co-location, stimulus intensity, abruptness, presentation rate, attentional allocation, and audiovisual meaningfulness affect responses. The sign convention for onset asynchrony should be explicit, and fitted parameters should be accompanied by uncertainty and observed response proportions. Apparatus calibration is especially important for comparisons across milliseconds. A temporal-order threshold is informative about a particular perceptual decision under controlled conditions; claiming that it equals the duration of a perceptual moment or the limit of awareness requires independent evidence that the procedure does not supply on its own Vroomen & Keetels (2010) O'Donohue et al. (2025).
The Multisensory Temporal Binding Window and Its Task Dependence
Auditory, visual, and tactile features from a common event often reach sensory receptors and cortical systems at different times. Perception can nevertheless integrate them into a single object or action-related occurrence. A multisensory temporal binding window denotes the temporal range over which a specified operational index indicates a greater probability of integration, common-cause attribution, or simultaneity. This range is learned and task-sensitive. Stevenson and Wallace (2013) directly addressed stimulus and task dependencies in multisensory temporal integration, establishing an important methodological challenge for treating any one window width as an invariant individual characteristic Stevenson & Wallace (2013).
Some studies estimate a binding interval by reporting the stimulus offsets that elicit synchrony decisions; others use the range over which one modality changes the report about another. These are different dependent variables. A perceptual illusion may persist at delays where an observer correctly judges two signals as sequential, because a common-event interpretation and an explicit temporal-order judgment use different evidence and criteria. The window can also be asymmetric, with broader tolerance for vision preceding sound under some circumstances. Such asymmetry should be interpreted in relation to natural event statistics, relative sensory precision, and decision demands rather than as a universal sensory constant Vroomen & Keetels (2010) Shams & Beierholm (2022).
The field has recently reexamined whether traditional measures converge. O'Donohue, Lacherez, and Yamamoto (2025) compared classical measures and the effects of experience, highlighting why task selection changes the apparent size of a multisensory integration window. Their work encourages comparisons of within-person correlations, learning, and parameter stability using designs that avoid conflating auditory-visual synchrony with response selection. A strong construct-validity claim would require multiple independent tasks, controlled stimulus differences, and a latent model accounting for measurement error. The availability of several related indices does not itself establish that they measure one underlying trait O'Donohue et al. (2025) Stevenson & Wallace (2013).
Crossmodal Illusions as Experimental Evidence of Temporal Integration
A striking demonstration of crossmodal interaction is the sound-induced flash illusion described by Shams, Kamitani, and Shimojo (2000). Brief auditory transients paired with a visual flash can alter the number of flashes reported, even when the physical visual input does not contain the reported events. This result demonstrates that auditory information can change visual event perception rather than merely accompanying it. It is valuable for temporal research because the effect depends on the timing and grouping of sensory signals, permitting controlled tests of when information from different modalities contributes to a single perceptual judgment Shams et al. (2000).
Temporal ventriloquism provides another example: auditory events can influence the apparent timing of visual events. Morein-Zamir, Soto-Faraco, and Kingstone (2003) experimentally investigated auditory capture of vision under temporal arrangements that altered visual timing judgments. These effects show that the sensory time attributed to an event may be reconstructed through crossmodal interactions. However, an altered response in a visual order task does not by itself determine whether the source is early perceptual capture, altered attention to markers, a changed decision criterion, or common-cause inference. Discriminating the alternatives requires manipulation of sensory reliability, irrelevant sound timing, and response instructions Morein-Zamir et al. (2003) García-Pérez & Alcalá-Quintana (2012).
Crossmodal illusions also reveal the danger of overgeneralization. Illusory flash reports and synchronization judgments capture different endpoints of integration. The presence of a dramatic illusion cannot establish a universal temporal window for speech, object perception, audiovisual navigation, and self-produced action. Conversely, a correct simultaneity report does not guarantee that all event features were perceptually fused. Strong analysis compares the conditions in which crossmodal input changes stimulus identity, order, duration, or causal attribution, and then tests whether a single probabilistic model predicts each effect out of sample Shams et al. (2000) Stevenson & Wallace (2013).
Temporal Recalibration and the Plasticity of Synchrony
Experienced sensory timing is adaptable. Fujisaki, Shimojo, Kashino, and Nishida (2004) demonstrated recalibration of audiovisual simultaneity following repeated exposure to an asynchrony. A stable delay between sound and image can change subsequent judgments of when the two modalities are simultaneous. This phenomenon is consistent with systems that compensate for persistent crossmodal delay, although it does not by itself identify whether adaptation occurs in early latency estimates, higher-level common-cause inference, or decision criteria. The design's importance is that temporal alignment behaves as a history-dependent estimate rather than a fixed physical zero point Fujisaki et al. (2004).
Attention modifies this adaptation. Heron and colleagues (2010) studied attention's role in the plasticity of multisensory timing, indicating that temporal recalibration is sensitive to the observer's processing of the adapting stimuli. A rigorous experiment must compare pre-adaptation and post-adaptation measurements under matched test conditions, include controls for fatigue and practice, and specify whether adaptation generalizes to new locations, features, or modalities. Bruns, Paumen, and Röder (2025) examined perceptual training and transfer of audiovisual simultaneity judgments across spatial locations. Such training research clarifies which aspects of timing plasticity are stimulus-specific and which support broader perceptual learning Heron et al. (2010) Bruns et al. (2025).
Recalibration should be distinguished from the temporal binding effects that occur within a single trial. Trial-by-trial sensory capture may reflect immediate inference using multiple cues, whereas adaptation changes the mapping from physical asynchrony to later perceptual report. Both can coexist, and both are sensitive to experiment design. Models that incorporate learning rates, statistical regularities, and uncertainty need independent measurements of the time course of recalibration and the persistence of its aftereffects. The distinction has implications for audiovisual communication systems, remote interaction, and augmented environments, where stable technical latencies may be perceptually accommodated to different degrees Fujisaki et al. (2004) Shams & Beierholm (2022).
Causal Inference, Common-Event Beliefs, and Statistical Integration
Whether two temporally nearby signals are combined depends on whether the observer treats them as arising from a common cause. Bayesian causal-inference models represent competing hypotheses: a shared source that should support integration, and separate sources for which combining estimates would introduce error. Shams and Beierholm (2022) presented the broader causal-inference approach as a unifying model for multisensory perception. Temporal proximity is informative but not decisive; spatial coincidence, semantic correspondence, reliability, and learned environmental regularities can influence the inferred causal relation Shams & Beierholm (2022).
Audiovisual speech illustrates why physical synchrony alone may be insufficient as an explanation. Facial movements and sound vary continuously and may support source identification even when modestly offset. For brief arbitrary flashes and beeps, the same temporal offset could yield a different report because the signals provide fewer coherent common-cause cues. A model should therefore estimate how correspondence changes the posterior probability of one event and should distinguish the evidence for common origin from the evidence about timing itself. Manipulations that degrade one modality can test predictions about reliability-weighted influence, although decision and attention effects require separate controls Stevenson & Wallace (2013) Vroomen & Keetels (2010).
Applying causal inference to time requires care about the target variable. An observer may integrate the timing of two sources, select the most reliable modality, tolerate misalignment while preserving event identity, or recalibrate the expected sensory delay. These computations have different error costs. It is methodologically useful to compare models through predictive checks on full distributions and conditions withheld from fitting. Flexible Bayesian stories can otherwise explain any post hoc direction of bias simply by adjusting priors. The strongest support arises from preregistered manipulations of physical synchrony, reliability, and common-cause cues with quantitatively different predictions Shams & Beierholm (2022) García-Pérez & Alcalá-Quintana (2012).
Temporal Integration in Visual Attention and Event Representation
Temporal integration occurs within a modality as well as across modalities. Rapidly successive visual elements may contribute jointly to an object representation, a feature estimate, or working-memory encoding. Akyürek (2025) reviewed temporal integration as an adaptive process at the interface of visual perception, attention, and working memory. The resulting conceptual picture treats information accumulation over time as useful for constructing stable perceptual estimates in a changing environment. The appropriate integration period depends on stimulus quality, task instructions, and what counts as one object or event Akyürek (2025).
Integrating over a longer interval can improve signal reliability by pooling information, yet can also obscure the order or identity of separate events. This tradeoff matters when the same display is used to measure apparent motion, event order, perceptual continuity, or conscious report. Sensory persistence, object updating, attention shifts, and decision thresholds may each contribute. Experiments need to specify whether the participant is reporting the features of one combined event, the order of two events, or the memory of a rapidly presented sequence. Treating all such results as a generic integration window loses these distinctions and weakens mechanistic inference Akyürek (2025) Samaha & Romei (2024).
Temporal integration is also subject to serial context. Perception on the current trial may be influenced by recently encountered durations, rhythms, or feature sequences. Such dependence can stabilize estimates in a noisy environment while producing systematic biases after a change in stimulus statistics. The interaction among attention, memory, and perceptual grouping suggests a research program that measures both immediate crossmodal binding and slower history-dependent adjustment. Its central methodological requirement is to separate the contribution of current signals from the influence of priors and preceding responses Jazayeri & Shadlen (2010) Akyürek (2025).
Neural Systems of Interval Timing: Distributed and Specialized Contributions
Research on neural timing identifies multiple interacting cortical and subcortical systems. Coull, Cheng, and Meck (2011) reviewed neuroanatomical and neurochemical substrates that contribute to interval timing and highlighted the roles of basal ganglia, frontal cortex, and other circuits under different task conditions. Merchant, Harrington, and Meck (2013) assembled evidence for distributed networks involved in perceiving and estimating time. These studies constrain anatomical claims: activation during a duration task may index time-related information, but it may also index attention, decision difficulty, action preparation, or memory. Assigning a region the status of a universal clock would require evidence of temporal coding and causal necessity across divergent paradigms Coull et al. (2011) Merchant et al. (2013).
Cortico-striatal accounts explain timing partly through evolving cortical activity sampled or integrated by basal-ganglia circuitry. Evidence from pharmacological, behavioral, lesion, and physiological studies suggests that dopaminergic and motor-related processes can influence temporal behavior, although each intervention has multiple effects. Changes in movement vigor, vigilance, reinforcement learning, and response threshold can resemble changes in clock speed. Quantitative neural studies therefore need matched control conditions and explicit response models. The critical question is whether the neural circuit predicts duration-specific error structure after the effects of non-temporal demands have been measured Buhusi & Meck (2005) Coull et al. (2011).
The contemporary synthesis emphasizes representational plurality. Paton and Buonomano (2018) reviewed mechanisms spanning population dynamics and specialized computations, with timing embedded in perception, learning, cognition, and action. A robust explanation can permit shared circuitry without requiring a single identical code for every temporal scale. Longitudinal neural recording, causal perturbation, and model-based analyses can establish whether a particular circuit contributes to the temporal variable itself or to the task in which that variable is used. Understanding time perception requires this linking of mechanism to observable behavior rather than a simple catalog of activated regions Paton & Buonomano (2018) Merchant et al. (2013).
Cerebellar Prediction, Sensory-Specific Dynamics, and Timing Across Scales
Cerebellar and sensorimotor mechanisms are often implicated in precisely timed movements and in predicting the consequences of actions. Their relevance to perceptual timing depends on the task and whether the measured outcome requires a motor response, an internal comparison, or sensory ordering. Mauk and Buonomano (2004) emphasized the diversity of neural timing computations, including mechanisms that emerge from network properties. Dedicated and intrinsic accounts therefore make different predictions about whether perturbing a circuit changes temporal precision selectively or affects broader processing. Interpretations should be conditioned on interval range, stimulus modality, training, and the controls included Mauk & Buonomano (2004) Ivry & Schlerf (2008).
A circuit that provides a stable temporal readout for one trained sequence may fail to generalize to an unrelated signal. This is informative rather than anomalous if the system codes temporal structure through task-dependent states. Karmarkar and Buonomano (2007) offer a concrete example of how a neural network can represent intervals through its evolving state. Testing these accounts requires comparing fixed and variable stimulus contexts, examining history effects, and distinguishing trained classification from general elapsed-time estimation. Large-scale representations may also be coordinated across systems, a possibility that remains compatible with distributed coding and limited specialization Karmarkar & Buonomano (2007) Paton & Buonomano (2018).
Cross-scale theories face a particular methodological difficulty. Millisecond sensory-order thresholds, subsecond interval discrimination, and seconds-long reproduction yield different noise sources and memory demands. A mechanistic result from one range may be suggestive but cannot be transported directly to another. Neuroimaging resolution, electrophysiological sampling, and causal intervention differ in what they can reveal. Stronger studies combine appropriately timed neural measurements with behavioral psychometrics and test whether neural parameters predict the distribution of judgments across manipulated intervals, not merely average accuracy at one duration Merchant et al. (2013) Grondin (2010).
Rhythm, Anticipation, and the Temporal Organization of Attention
Temporal structure enables attention to be allocated to moments when information is likely to occur. Nobre and van Ede (2018) reviewed how learned temporal expectations shape anticipatory selection. An observer receiving regular cues may improve processing at expected positions in a sequence; a probabilistic temporal cue can similarly change readiness for a stimulus without requiring a subjective duration judgment. These effects connect time perception to selective attention, but expectancy, response preparation, and perceptual accuracy must be measured separately. An improvement in response time at a predicted moment may arise from perceptual facilitation, decision readiness, or motor preparation Nobre & van Ede (2018).
Rhythmic stimulation also permits analysis of neural phase and entrainment. Samaha and Romei (2024) reviewed evidence associating alpha-band frequency with temporal windows in perception, including a living meta-analytic component. Such research addresses whether the temporal segmentation of incoming information covaries with features of brain oscillations. Correlations between individual alpha frequency and behavioral thresholds can motivate mechanistic hypotheses, yet they do not establish that each oscillatory cycle constitutes a conscious perceptual frame. Experimental manipulations of rhythm, task demands, and phase-dependent performance are needed to test competing interpretations Samaha & Romei (2024).
Temporal prediction need not rely on a periodic oscillator. A learned hazard function—the changing probability of an event conditional on its not yet occurring—can structure attention during irregular sequences. A participant's preparedness can therefore change as elapsed time increases, even when events have no fixed beat. Experimental analyses should distinguish rhythmic phase effects from conditional-probability learning and strategic waiting. The resulting framework links interval expectation to attention and action while preserving the separate questions of how time is represented and how temporal information changes task performance Nobre & van Ede (2018) Jazayeri & Shadlen (2010).
Sensorimotor Synchronization, Agency, and Action–Effect Timing
Sensorimotor synchronization provides a naturalistic window into temporal prediction. When a person taps along with a metronome, the resulting sequence of action–sound asynchronies reveals anticipatory control and error correction. Repp and Su (2013) reviewed extensive studies of synchronization and demonstrated the relevance of phase correction, tempo adaptation, and motor variability. A participant may maintain a stable average tempo while continually adjusting the timing of individual movements, so mean interval reproduction alone would conceal important control dynamics. Musical experience and sensory modality can influence these dynamics, making calibration and task description essential Repp & Su (2013).
Self-generated actions alter the interpretation of temporal proximity. Haggard, Clark, and Kalogeras (2002) studied the perceived temporal relation between voluntary action and its effects, contributing to the literature on intentional binding. Stetson and colleagues (2006) showed that sensorimotor recalibration can produce striking changes in the perceived order of action and sensation. These paradigms illuminate interactions among agency attribution, causal prediction, sensory feedback, and timing. They also require caution: a contraction in reported action–effect intervals may arise from multiple cognitive and perceptual mechanisms and cannot, by itself, establish a unique internal sense of agency Haggard et al. (2002) Stetson et al. (2006).
Action–effect timing and passive audiovisual synchrony provide a useful comparison. In the action condition, an observer may predict sensory consequences using motor commands and learned contingencies; passive conditions lack the same efferent information. Comparing the two while matching sensory signals can test whether motor predictions contribute specifically to perceived event order or to a more general decision criterion. Changes after delayed feedback can reflect recalibration of sensory timing, of motor timing, or of the causal expectation linking them. Integrated models should predict both behavioral adaptation and the distribution of explicit judgments Stetson et al. (2006) Fujisaki et al. (2004).
Development, Individual Differences, and Clinical-Adjacent Evidence
Temporal abilities develop with sensory maturation, attention, motor control, and memory. Comparisons across age groups must take account of differing task comprehension, motor response variability, and experience with the specific stimulus modality. A child's broad temporal-order function could reflect sensory noise, inconsistent response mapping, or greater lapse rates rather than a uniquely wider neural binding window. Similarly, expertise with musical timing or audiovisual events can change performance through training and prior exposure. These considerations make within-person learning and reliability assessments especially valuable in developmental research Merchant et al. (2013) O'Donohue et al. (2025).
Research comparing temporal processing across neurodevelopmental populations illustrates both potential significance and methodological difficulty. Meilleur and colleagues (2020) reviewed and meta-analyzed unisensory and multisensory timing findings in autism and dyslexia. Such syntheses can reveal patterns across heterogeneous studies, but pooled effects depend on the tasks selected, age groups, diagnostic ascertainment, measurement reliability, and publication practices. A difference in a laboratory measure cannot by itself serve as a diagnosis or establish that temporal integration causes difficulties in language, learning, or social communication. Appropriate interpretation requires convergent measures, longitudinal designs, and formal consideration of competing developmental processes Meilleur et al. (2020).
Within-person variability is also scientifically meaningful. Sleep, attention, motivation, sensory acuity, and practice can change timing behavior between sessions. Recent studies of multisensory windows and the latent structure of duration perception sharpen the question of whether individuals display stable timing factors across several paradigms or a collection of task-specific abilities. A rigorous individual-difference program should measure retest reliability, sensory-motor confounds, and model-based uncertainty before interpreting correlations. The goal is to determine which properties persist across tasks and contexts and what they predict beyond the measurement occasion Sadibolova et al. (2025) O'Donohue et al. (2025).
Temporal Dynamics in Artificial Systems and Human–Artificial Coordination
Artificial systems provide a useful comparative domain for testing functional timing architectures. A perception system can timestamp incoming signals, compensate for sensor latencies, estimate asynchrony, fuse observations over adaptive windows, and predict future events. These functions are operationally comparable with selected human timing tasks when the input statistics, constraints, response variable, and evaluation criteria are matched. A deterministic timestamp is a feature of an engineered system, whereas human subjective duration is inferred from behavior; their comparison must specify the functional problem under study. The scientific advantage is that artificial implementations can make assumptions about clocks, priors, fusion rules, and memory explicit and experimentally manipulable Paton & Buonomano (2018) Shams & Beierholm (2022).
In human–artificial interaction, timing influences turn-taking, shared attention, conversational interruption, audiovisual correspondence, and the interpretation of contingent action. Delayed speech responses or mismatched video can change whether communication feels coordinated. The relevant psychological hypothesis is that people use temporal contingencies as evidence for event unity, agency, and interpersonal responsiveness. Testing it requires randomizing latency, measuring behavior and attribution independently, and controlling content quality. A software system's apparent social timing is therefore a measurable interaction property whose consequences depend on the human observer's expectations and the system's behavior Nobre & van Ede (2018) Haggard et al. (2002).
Within the Artificial Era horizon, temporal organization becomes an explicit comparative research question across biological and nonbiological agents. Scientific claims should identify observable performance: sequence learning, event-order inference, multimodal synchronization, adaptive delay compensation, and long-horizon continuity. These methods can evaluate functional capacities without treating an engineered timing mechanism as evidence of subjective experience. The next scientific step is preregistered cross-system comparisons using transparent timing distributions and shared measurement criteria, while retaining the distinct provenance of human psychological evidence and artificial-system engineering results Ivry & Schlerf (2008) Akyürek (2025).
Critical Evaluation of Behavioral Evidence and Measurement Validity
Timing paradigms are often described through compact indices: a Weber fraction, a reproduced interval, a simultaneity threshold, a point of subjective simultaneity, or a synchronization error. Each index is informative only with respect to the procedure that generates it. A single duration-reproduction mean confounds encoding, memory, motor execution, and response criterion. A temporal-order judgment can reflect perceptual-latency differences and decision mapping. A simultaneity function can change when the participant adopts a stricter definition of 'same event.' The central issue is construct validity: what psychological process is necessary to explain the measured change, and what alternative route could produce an equivalent response pattern? García-Pérez & Alcalá-Quintana (2012) Grondin (2010).
Strong experimental designs create conditions under which explanations diverge. Varying uncertainty independently of interval range tests probabilistic calibration; varying temporal relevance separately from perceptual salience helps distinguish attentional allocation; crossmodal control conditions identify whether a timing effect generalizes beyond one sensory pairing. Distributions and confidence intervals matter because changes in response variance, bias, and lapse rate can oppose changes in point estimates. Preregistered contrast definitions, access to trial-level data, documented exclusions, and transparent fitting improve the interpretability of these experiments. A nominally successful replication should reproduce the operational effect and its uncertainty, not merely a similar verbal interpretation Jazayeri & Shadlen (2010) Stevenson & Wallace (2013).
Cross-task comparisons require particular caution. A person may perform consistently in rhythmic prediction while showing weak reliability on a sparse temporal-order test. Correlations may be attenuated by measurement noise, while artificial relationships may result from shared response strategies and general cognitive ability. Latent-variable models can help when tasks have sufficient reliability and distinct sources of error, but factor interpretations depend on the measurements included. The recent search for the latent structure of time perception is valuable precisely because multiple observed tasks are needed to test whether a global temporal capacity is scientifically warranted Sadibolova et al. (2025) O'Donohue et al. (2025).
Neural Claims, Causal Identification, and Replicability
A neural population can represent elapsed time as an evolving trajectory while also representing the stimulus, expected response, and task state. This multidimensionality complicates claims that a recorded firing pattern implements a dedicated clock. Perturbing a circuit can strengthen causal inference, but altered timing behavior can still arise indirectly through motivation, attention, movement, or learning. The most convincing tests combine temporally precise recordings, selective interventions, and explicit computational models that predict psychometric outcomes. The neural evidence is strongest when it survives changes in task format and when responses to nontemporal control conditions are preserved Karmarkar & Buonomano (2007) Paton & Buonomano (2018).
Reviews of timing neurobiology provide convergent candidates rather than a single final architecture. Cortico-striatal mechanisms, cerebellar computations, recurrent cortical states, and sensory-specific dynamics have each been implicated under different conditions. An anatomical generalization should specify interval range, modality, motor demands, and clinical or experimental population. Nor should pharmacological effects be interpreted as isolated changes to an internal pacemaker without ruling out changes in decision and action. Model comparison at the computational level can clarify which part of a behavioral change is attributed to timing and which to broader processing Buhusi & Meck (2005) Coull et al. (2011).
Replicability requires sensitivity to operational differences across laboratories. Screen and audio latency, psychometric fitting conventions, duration ranges, adaptation schedules, and participant instructions are all capable of shifting measured temporal thresholds. Researchers comparing studies should record these choices and assess risk of selective reporting. A literature with different results can contain meaningful boundary conditions as well as errors. Replication and synthesis benefit from open data, shared stimulus-calibration documentation, explicit priors, negative findings, and quantitative rather than solely narrative comparisons of competing models Samaha & Romei (2024) O'Donohue et al. (2025).
Recent Research: Integration, Latent Structure, and Learning
Recent work has sharpened how temporal integration is conceptualized and measured. Akyürek (2025) reviewed integration across visual perception, attention, and working memory, emphasizing the adaptive accumulation of information rather than a single window with one fixed length. That perspective suggests experiments that vary uncertainty and task relevance while determining when temporally separated features are assigned to one representation. It also offers an opportunity to test whether integration precision and duration estimates share constrained forms of statistical inference or use distinct representational operations Akyürek (2025).
Sadibolova and colleagues (2025) investigated the latent structure of human time perception using multiple behavioral measures. The central importance of this program is methodological: task-specific noise and strategy can produce patterns that resemble separate capacities or mask common ones. Large and heterogeneous batteries, test–retest measurement, and independent validation are needed to establish stable latent constructs. An identified factor should earn its name from the tasks that support it, with confidence in its generalization proportional to cross-sample replication. The study contributes to a more cautious description of timing as a potentially multidimensional psychological field Sadibolova et al. (2025).
Training and recalibration remain active areas of investigation. O'Donohue and colleagues (2025) compared experience-related effects on measures of multisensory temporal integration, while Bruns and colleagues (2025) investigated transfer of audiovisual simultaneity training across space. These studies motivate sharper distinctions among rapid recalibration, sustained learning, and stable individual differences. The scientific question is not simply whether a threshold improves with practice; it is whether learning changes sensory precision, a prior expectation, a decision criterion, or an internal causal model, and whether that change predicts performance on new stimuli and tasks O'Donohue et al. (2025) Bruns et al. (2025).
Open Scientific Problems and a Research Agenda
One open problem is the degree of computational unification across timing tasks. Scalar expectancy theory explains prominent distributional properties, Bayesian approaches capture adaptation to contextual statistics, and neural-state theories demonstrate how intervals can be encoded in evolving activity. Whether these explanations occupy different functional levels of the same system or capture fundamentally different timing operations requires experiments that measure the same participants across controlled timing regimes. A successful unifying account would predict when parameters transfer and when task-specific representations are needed, rather than presuming that every interval judgment reads one clock Gibbon (1977) Jazayeri & Shadlen (2010) Ivry & Schlerf (2008).
A second challenge concerns the measurement of temporal integration. Simultaneity, order, flash illusions, temporal ventriloquism, and action–effect binding yield different signatures of temporal combination. Research needs designs that measure several signatures under shared stimulus conditions and test models of sensory latency, causal attribution, and response criterion jointly. It also needs attention to asymmetric offsets, modality reliability, and the adaptation history of each participant. Only such comparisons can establish when a reported temporal binding window represents a stable latent computation rather than a convenient index for one paradigm García-Pérez & Alcalá-Quintana (2012) Shams & Beierholm (2022).
A third challenge is ecological and developmental generalization. Natural events have correlated spatial, temporal, and semantic structure, whereas laboratory tasks often use brief artificial signals. Neural signals recorded in tightly constrained settings may not generalize to complex interaction. Across ages and populations, task comprehension and response demands can obscure genuine sensory differences. Future work should combine controlled laboratory psychophysics with calibrated naturalistic stimuli, preregistered prediction, longitudinal observations, and computational simulations whose timing assumptions are transparent. Comparative artificial systems may help test functional principles when matching criteria are established, while biological claims remain anchored in appropriate human or animal evidence Meilleur et al. (2020) Paton & Buonomano (2018).
Integrated Scientific Synthesis
The evidence supports several connected propositions. Temporal discrimination is a measurement problem with identifiable noise and decision components; contextual regularities influence estimates; attention and memory alter the evidence used for duration judgments; and multisensory event inference tolerates and adapts to physical delays. Neural timing is implemented through interacting circuits and evolving population activity whose contributions depend on task and timescale. These propositions have convergent support from formal models, psychophysics, learning experiments, neural research, and multisensory demonstrations. Their synthesis does not imply that one mechanism alone produces every form of temporal behavior Grondin (2010) Coull et al. (2011) Shams & Beierholm (2022).
Duration, sequence, simultaneity, and integration are best evaluated as a family of linked but analytically distinct functions. Duration estimates refer to interval magnitude; order judgments depend on relative event timing; simultaneity judgments concern the criterion and evidence for co-occurrence; temporal integration determines whether information accumulated over time contributes to one representation or response. Learning can adjust all of them, but the observed pattern depends on stimulus statistics and the organism's task. A shared vocabulary of precision, prior probability, decision criterion, integration, and motor correction permits productive comparison without erasing experimental distinctions García-Pérez & Alcalá-Quintana (2012) Jazayeri & Shadlen (2010) Repp & Su (2013).
The central methodological consequence is straightforward. Scientific claims about time perception gain strength when they are grounded in calibrated stimuli, explicit operational definitions, competing models, full response distributions, and convergent causal evidence. A theory should specify what it predicts before the outcome is known and which plausible non-temporal process has been controlled. Historical clock models remain important formal tools; newer accounts of distributed activity, context-sensitive inference, and temporal grouping expand the testable explanation. Their combined research program situates human temporal experience within perception, cognition, action, and the evolving study of coordinated biological and artificial systems Gibbon et al. (1984) Paton & Buonomano (2018) Akyürek (2025).
Conclusion
Time Perception and Temporal Integration constitute an interconnected research domain concerned with how events are timed, ordered, compared, predicted, and assigned to unified representations. Its scientific history moves from classical interval discrimination and internal-clock theory toward formal scalar models, attention and memory analyses, probabilistic inference, multisensory psychophysics, and neural population dynamics. Each tradition contributes distinct explanatory constraints. The cumulative evidence shows that duration judgments, temporal-order decisions, simultaneity reports, crossmodal binding, and action synchronization can vary partly independently and are shaped by uncertainty and prior experience Treisman (1963) Gibbon (1977) Merchant et al. (2013).
The strongest synthesis identifies multiple temporal computations interacting with broader perceptual and cognitive processes. Evidence for scalar variability does not uniquely specify a pacemaker; a shift in perceived synchrony does not uniquely reveal a sensory delay; and a neural timing correlate does not establish causal implementation without selective tests. Temporal adaptation, predictive attention, and action–effect coordination establish the flexibility and practical significance of temporal organization. Contemporary investigations of latent structure and temporal learning are especially important because they test which distinctions reflect stable cognitive properties and which arise from particular measurement choices Fujisaki et al. (2004) Nobre & van Ede (2018) Sadibolova et al. (2025).
The outstanding scientific program is to explain when the brain maintains separate temporal representations, when they are integrated into shared event estimates, and how those computations generalize across modalities, behavior, and development. Progress depends on rigorous psychophysics, accessible empirical evidence, model discrimination, and calibrated comparisons across contexts. For Psychology by Bogdanova, the result is a fundamental account of temporal organization within Perception and Psychophysics, connected to cognitive and neural science and open to carefully operationalized study of human–artificial coordination in the Artificial Era Paton & Buonomano (2018) Akyürek (2025).
References
- Akyürek, E. G. (2025). Temporal integration as an adaptive process in visual perception, attention, and working memory. Neuroscience & Biobehavioral Reviews, 170, 106041. https://doi.org/10.1016/j.neubiorev.2025.106041
- Block, R. A., & Zakay, D. (1997). Prospective and retrospective duration judgments: A meta-analytic review. Psychonomic Bulletin & Review, 4(2), 184–197. https://doi.org/10.3758/BF03209393
- Brown, S. W. (1997). Attentional resources in timing: Interference effects in concurrent temporal and nontemporal working memory tasks. Perception & Psychophysics, 59(7), 1118–1140. https://doi.org/10.3758/BF03205526
- Bruns, P., Paumen, T., & Röder, B. (2025). Perceptual training of audiovisual simultaneity judgments generalizes across spatial locations. Perception, 54(8), 609–627. https://doi.org/10.1177/03010066251342010
- Buhusi, C. V., & Meck, W. H. (2005). What makes us tick? Functional and neural mechanisms of interval timing. Nature Reviews Neuroscience, 6, 755–765. https://doi.org/10.1038/nrn1764
- Church, R. M. (1984). Properties of the internal clock. Annals of the New York Academy of Sciences, 423, 566–582. https://doi.org/10.1111/j.1749-6632.1984.tb23459.x
- Coull, J. T., Cheng, R.-K., & Meck, W. H. (2011). Neuroanatomical and neurochemical substrates of timing. Neuropsychopharmacology, 36, 3–25. https://doi.org/10.1038/npp.2010.113
- Eagleman, D. M., Tse, P. U., Buonomano, D., Janssen, P., Nobre, A. C., & Holcombe, A. O. (2005). Time and the brain: How subjective time relates to neural time. Journal of Neuroscience, 25(45), 10369–10371. https://doi.org/10.1523/JNEUROSCI.3487-05.2005
- Fujisaki, W., Shimojo, S., Kashino, M., & Nishida, S. (2004). Recalibration of audiovisual simultaneity. Nature Neuroscience, 7, 773–778. https://doi.org/10.1038/nn1268
- García-Pérez, M. A., & Alcalá-Quintana, R. (2012). On the discrepant results in synchrony judgment and temporal-order judgment tasks: A quantitative model. Psychonomic Bulletin & Review, 19(5), 820–846. https://doi.org/10.3758/s13423-012-0278-y
- Gibbon, J. (1977). Scalar expectancy theory and Weber's law in animal timing. Psychological Review, 84(3), 279–325. https://doi.org/10.1037/0033-295X.84.3.279
- Gibbon, J., Church, R. M., & Meck, W. H. (1984). Scalar timing in memory. Annals of the New York Academy of Sciences, 423, 52–77. https://doi.org/10.1111/j.1749-6632.1984.tb23417.x
- Grondin, S. (2010). Timing and time perception: A review of recent behavioral and neuroscience findings and theoretical directions. Attention, Perception, & Psychophysics, 72(3), 561–582. https://doi.org/10.3758/APP.72.3.561
- Haggard, P., Clark, S., & Kalogeras, J. (2002). Voluntary action and conscious awareness. Nature Neuroscience, 5(4), 382–385. https://doi.org/10.1038/nn827
- Heron, J., Roach, N. W., Whitaker, D., & Hanson, J. V. M. (2010). Attention regulates the plasticity of multisensory timing. European Journal of Neuroscience, 31(10), 1755–1762. https://doi.org/10.1111/j.1460-9568.2010.07194.x
- Ivry, R. B., & Schlerf, J. E. (2008). Dedicated and intrinsic models of time perception. Trends in Cognitive Sciences, 12(7), 273–280. https://doi.org/10.1016/j.tics.2008.04.002
- Jazayeri, M., & Shadlen, M. N. (2010). Temporal context calibrates interval timing. Nature Neuroscience, 13, 1020–1026. https://doi.org/10.1038/nn.2590
- Kanai, R., Paffen, C. L. E., Hogendoorn, H., & Verstraten, F. A. J. (2006). Time dilation in dynamic visual display. Journal of Vision, 6(12), 1421–1430. https://doi.org/10.1167/6.12.8
- Karmarkar, U. R., & Buonomano, D. V. (2007). Timing in the absence of clocks: Encoding time in neural network states. Neuron, 53(3), 427–438. https://doi.org/10.1016/j.neuron.2007.01.006
- Mauk, M. D., & Buonomano, D. V. (2004). The neural basis of temporal processing. Annual Review of Neuroscience, 27, 307–340. https://doi.org/10.1146/annurev.neuro.27.070203.144247
- Meilleur, A., Foster, N. E. V., Coll, S.-M., Brambati, S. M., & Hyde, K. L. (2020). Unisensory and multisensory temporal processing in autism and dyslexia: A systematic review and meta-analysis. Neuroscience & Biobehavioral Reviews, 116, 44–63. https://doi.org/10.1016/j.neubiorev.2020.06.013
- Merchant, H., Harrington, D. L., & Meck, W. H. (2013). Neural basis of the perception and estimation of time. Annual Review of Neuroscience, 36, 313–336. https://doi.org/10.1146/annurev-neuro-062012-170349
- Morein-Zamir, S., Soto-Faraco, S., & Kingstone, A. (2003). Auditory capture of vision: Examining temporal ventriloquism. Cognitive Brain Research, 17(1), 154–163. https://doi.org/10.1016/S0926-6410(03)00089-2
- Nobre, A. C., & van Ede, F. (2018). Anticipated moments: Temporal structure in attention. Nature Reviews Neuroscience, 19, 34–48. https://doi.org/10.1038/nrn.2017.141
- O'Donohue, M. P., Lacherez, P., & Yamamoto, N. (2025). Effects of short- and long-term experience on two classical measures of the multisensory temporal integration window. Journal of Experimental Psychology: Human Perception and Performance, 51(3), 386–404. https://doi.org/10.1037/xhp0001278
- Pariyadath, V., & Eagleman, D. (2007). The effect of predictability on subjective duration. PLOS ONE, 2(11), e1264. https://doi.org/10.1371/journal.pone.0001264
- Paton, J. J., & Buonomano, D. V. (2018). The neural basis of timing: Distributed mechanisms for diverse functions. Neuron, 98(4), 687–705. https://doi.org/10.1016/j.neuron.2018.03.045
- Repp, B. H., & Su, Y.-H. (2013). Sensorimotor synchronization: A review of recent research (2006–2012). Psychonomic Bulletin & Review, 20(3), 403–452. https://doi.org/10.3758/s13423-012-0371-2
- Sadibolova, R., Widmer, C., Fletcher, Z., Weill, S., & Terhune, D. B. (2025). Uncovering the latent structure of human time perception. Cognition, 257, 106078. https://doi.org/10.1016/j.cognition.2025.106078
- Samaha, J., & Romei, V. (2024). Alpha-band frequency and temporal windows in perception: A review and living meta-analysis of 27 experiments (and counting). Journal of Cognitive Neuroscience, 36(4), 640–654. https://doi.org/10.1162/jocn_a_02069
- Shams, L., Kamitani, Y., & Shimojo, S. (2000). What you see is what you hear. Nature, 408, 788. https://doi.org/10.1038/35048669
- Shams, L., & Beierholm, U. (2022). Bayesian causal inference: A unifying neuroscience theory. Neuroscience & Biobehavioral Reviews, 137, 104619. https://doi.org/10.1016/j.neubiorev.2022.104619
- Stetson, C., Cui, X., Montague, P. R., & Eagleman, D. M. (2006). Motor-sensory recalibration leads to an illusory reversal of action and sensation. Neuron, 51(5), 651–659. https://doi.org/10.1016/j.neuron.2006.08.006
- Stevenson, R. A., & Wallace, M. T. (2013). Multisensory temporal integration: Task and stimulus dependencies. Experimental Brain Research, 227(2), 249–261. https://doi.org/10.1007/s00221-013-3507-3
- Treisman, M. (1963). Temporal discrimination and the indifference interval: Implications for a model of the internal clock. Psychological Monographs: General and Applied, 77(13), 1–31. https://doi.org/10.1037/h0093864
- Tse, P. U., Intriligator, J., Rivest, J., & Cavanagh, P. (2004). Attention and the subjective expansion of time. Perception & Psychophysics, 66(7), 1171–1189. https://doi.org/10.3758/BF03196844
- Vroomen, J., & Keetels, M. (2010). Perception of intersensory synchrony: A tutorial review. Attention, Perception, & Psychophysics, 72(4), 871–884. https://doi.org/10.3758/APP.72.4.871
- Wearden, J. H. (1991). Do humans possess an internal clock with scalar timing properties? Learning and Motivation, 22(1–2), 59–83. https://doi.org/10.1016/0023-9690(91)90017-3
- Zakay, D., & Block, R. A. (2004). Prospective and retrospective duration judgments: An executive-control perspective. Acta Neurobiologiae Experimentalis, 64(3), 319–328. https://doi.org/10.55782/ane-2004-1516