Skip to main content
NCBI home page
Wiley Open Access Collection logoLink to Wiley Open Access Collection
. 2026 Apr 3;117(4):1396–1410. doi: 10.1111/bjop.70070

Serial dependence in time perception requires consistent motor responses, not shared memory alone

Jiao Wu 1,✉, Halid Oğuz Serçe 2, Zhuanghua Shi 1
PMCID: PMC13655987  PMID: 41933500

Abstract

Serial dependence—the bias from recent experience on present response—has been attributed to shared memory representations, yet previous studies yielded contradictory findings about whether consistent motor responses are required. To address this debate in time perception, we tested whether serial dependence emerges when tasks share identical stimulus features but differ only in response modes. We interleaved temporal reproduction and bisection tasks using a post‐cue design that held duration encoding constant while varying motor output across trials. Grounded in binding and retrieval in action control (BRAC) theory, we hypothesized that response‐feature binding retrieval drives serial dependence. Using structural equation modelling (SEM), we dissociated perceptual (stimulus‐driven) and decisional (response‐driven) components. With the same task, we replicated repulsive perceptual serial dependence and attractive decisional carryover. Critically, both effects vanished across tasks despite identical stimulus processing—demonstrating that response‐mode consistency, not shared memory alone, drives sequential biases in temporal judgements. Our SEM approach uncovered repulsive perceptual influences that standard regression missed, highlighting its power to isolate overlapping effects. These findings reveal that response‐specific reactivation of event files underpins serial dependence in temporal decision‐making.

Keywords: decisional carryover, motor response, sequential biases, serial dependence

BACKGROUND

Imagine trying to judge whether a line you see is vertical or slightly tilted after viewing a series of tilted lines. Your judgement tends to lean towards the orientation you have most recently observed, reflecting a systematic bias caused by prior experience. This bias, known as serial dependence or sequential effect (Fischer & Whitney, 2014; Manassi et al., 2023; Pascucci et al., 2023), reveals how recent sensory history shapes current perceptual decisions. Serial dependence has been documented across various stimulus properties, including orientation, motion direction, position, colour, shape and numerosity (Cicchini et al., 2024; Pascucci et al., 2023). Crucially, such effects can occur at multiple cognitive levels, from early perceptual mechanisms (Burr & Cicchini, 2014; Cicchini et al., 2017, 2021; Fischer & Whitney, 2014; Kiyonaga et al., 2017; Liberman et al., 2016; Manassi et al., 2018; Murai & Whitney, 2021) to high‐level functions involving memory and decision‐making (Bliss et al., 2017; Feigin, Baror, et al., 2021; Fischer et al., 2024; Fritsche et al., 2017; Luo et al., 2025; Markov et al., 2024; Pascucci et al., 2019; Zhang & Lewis‐Peacock, 2024).

Recent research has decomposed the observed serial dependence into perceptual repulsion and decisional attraction (Hahn & Wei, 2024; Wei & Stocker, 2015), identifying these two contrasting dependencies by isolating the impacts of prior stimulus and response (Pascucci et al., 2019; Sadil et al., 2024). Various analytical approaches—including statistical modelling, computational generative models and specialized task designs—have successfully revealed these opposing effects (Ceylan & Pascucci, 2023; Fritsche et al., 2017; Gallagher & Benton, 2024; Pascucci & Plomp, 2021). This work suggests that early stage repulsion may be overridden at post‐perceptual stages, leading to the eventual observation of an attractive serial dependence (Hajonides et al., 2023; Sheehan & Serences, 2022).

Time perception, like orientation or colour, is also subject to serial dependence (Cheng, Chen, Glasauer, et al., 2024; Glasauer & Shi, 2022; Li et al., 2023; Wang et al., 2023; Wiener et al., 2014). Unlike orientation or colour, perceived duration lacks a dedicated sensory system and instead relies on distributed neural and cognitive processes (Ivry & Schlerf, 2008; Merchant et al., 2013). Interestingly, sequential biases have been observed across various timing tasks, including duration reproduction and temporal bisection, even when the classical central tendency effect was carefully controlled for (Bertolasi et al., 2025; Cheng, Chen, & Shi, 2024; Glasauer & Shi, 2022; Li et al., 2023). Intriguingly, however, magnitudes of sequential effects in timing depend on task context or on switching between modalities or tasks (Bae & Luck, 2020; Cheng, Chen, & Shi, 2024; Cheng, Chen, Yang, & Shi, 2024; Li et al., 2023). For instance, the sequential effect was reduced or diminished when the task was switched across trials, requiring participants to reproduce either direction or duration based on a trial‐by‐trial cue (Cheng, Chen, & Shi, 2024).

These modality‐ and task‐specific sequential effects suggest that post‐perceptual processes, particularly working memory, play a key role in serial dependence (Bae & Luck, 2020; Bliss et al., 2017; Cheng, Chen, Yang, & Shi, 2024). Maintaining the same task‐relevant information across trials (e.g. duration) can facilitate the integration of past and present stimuli in working memory and/or reactivation of subthreshold traces of recent memories (Bae & Luck, 2019; Barbosa & Compte, 2020), resulting in an attraction bias (Cheng, Chen, Glasauer, et al., 2024; Cheng, Chen, & Shi, 2024; Cheng, Chen, Yang, & Shi, 2024). Yet most findings on serial dependence related to task‐specific or modality‐specific effects come from studies interleaving tasks with distinct features or modalities, such as direction versus time judgements (Cheng, Chen, & Shi, 2024; Cheng, Chen, Yang, & Shi, 2024), numerosity versus time judgements (Fornaciai et al., 2023; Togoli et al., 2021), or colour versus direction tasks (Bae & Luck, 2020). Such distinct task pairings typically require participants to concurrently encode two separate stimulus attributes (e.g. motion direction and duration) into working memory. Selecting one task typically enhances the representation of its relevant feature while suppressing those of its irrelevant features, resulting in minimal or no serial dependence across tasks (e.g. Bae & Luck, 2020).

This leaves a critical gap: does serial dependence persist across interleaved tasks that share identical stimulus features but differ only in response requirements? The answer may hinge on whether serial dependence is rooted in perceptual or decisional processes, an issue the broader literature has yet to resolve clearly. In orientation and face perception, serial dependence can emerge without prior responses and transfer across different task formats (Fischer & Whitney, 2014; Liberman et al., 2016; Manassi & Whitney, 2022; Murai & Whitney, 2021), suggesting a perceptual component that operates independently of response mode. Pascucci et al. (2019), however, found that explicitly withholding responses and decisions eliminated attractive biases entirely, with repulsive effects emerging in their place, pointing instead to a decisional origin. In timing, the evidence leans more clearly towards decisions: serial dependence in duration reproduction did not survive without prior reports (Chen et al., 2023; Cheng, Chen, Glasauer, et al., 2024; Wang et al., 2023), and perceptual decisions on timing are systematically drawn towards prior responses—a phenomenon termed decisional carryover (Cheng, Chen, Yang, & Shi, 2024; Li et al., 2023; Wehrman et al., 2023; Wehrman et al., 2020; Wiener et al., 2014). This sharper response‐dependence in timing may reflect the tight coupling between temporal processing and motor circuits, making temporal representations more sensitive to the response context than representations in domains served by more autonomous sensory pathways. These considerations raise a more specific question: given that serial dependence operates at multiple processing levels, does the balance between perceptual and decisional components shift when response modes change—particularly in the timing domain, where stimulus representations may be more tightly coupled to motor processes? Specifically, if perceptual‐level mechanisms suffice, interleaving two timing tasks with identical stimulus features should yield comparable sequential biases despite different response modes. If, however, response‐feature binding plays a critical role, identical stimulus content may yield attenuated or absent serial dependence when response modes switch.

We addressed this question by grounding our predictions in the binding and retrieval in action control (BRAC) theory (Frings et al., 2020), which proposes that stimulus features and responses are bound together as event files in working memory. According to BRAC, retrieving the same response‐mode reactivates the whole event file, including stimulus features (Allenmark et al., 2025; Hommel, 2004). We hypothesized that this response‐feature binding mechanism drives serial dependence: when the same response mode is repeated across trials (e.g. reproduction → reproduction), reactivating that motor response triggers reactivation of stimulus features from previous trials, producing stronger serial dependence. Conversely, when response modes switch (e.g. reproduction → bisection), different motor responses will not trigger reactivation of previous stimulus features, resulting in weaker or absent cross‐task serial dependence. This framework can account for the previously contradictory findings in timing. In Fischer and Whitney (2014), the response mode (adjustment) was maintained throughout the experiment, even on no‐response trials where participants likely prepared the default motor response. This preserved response‐feature binding could explain why serial dependence persisted. In contrast, when responses were explicitly withheld and suppressed (Pascucci et al., 2019), the response mode changed effectively, disrupting binding and leading to diminished attractive serial dependence.

To directly test whether shared stimulus features alone drive serial dependence or whether response‐mode consistency is critical, we employed a methodological innovation that previous studies lacked: systematic task‐switching while maintaining identical stimulus processing. Unlike Fischer and Whitney (2014), who used a single task type (adjustment) with occasional no‐response trials, or Pascucci et al. (2019), who contrasted response versus no‐response conditions, we interleaved two tasks requiring fundamentally different response modes—duration reproduction (continuous motor adjustment) and temporal bisection (binary categorical choice)—while keeping stimulus features (duration) identical.

Participants performed trials involving either reproducing a given interval by manual key‐pressing (reproduction task) or categorizing intervals as short or long by binary responses (bisection task). Critically, we employed a post‐cue design: participants first encoded and maintained the duration, then executed the required task only after a cue was presented, ensuring identical perceptual encoding and memory maintenance across trials. This design isolates potential contributions of motor response from those related to perceptual and memory encoding. By examining serial dependence both within tasks (same response mode) and across tasks (different response modes), we could directly test whether response‐mode consistency, rather than shared memory encoding alone, drives sequential biases in time perception.

METHODS

Participants

We recruited 24 volunteers (16 females, 28.5 ± 4.49 years) for the experiment. The sample size was based on effect sizes, ranging from 0.596 to 1.84, reported in previous studies on serial dependence in duration reproduction tasks (Cheng, Chen, Yang, & Shi, 2024; Glasauer & Shi, 2022) and the effect size of 0.684 in a duration bisection task (Cheng, Chen, Yang, & Shi, 2024). To ensure sufficient power (80%) at an alpha level of 0.05, we used the smallest effect size (0.596), which yielded at least 19 participants according to G*Power calculation (Erdfelder et al., 1996). To be conservative, we recruited 24 participants to assess both task‐relevant and potential task‐irrelevant serial dependence. All reported normal or corrected normal vision, gave informed consent, and received monetary or course credit compensation for their participation. The Ethics Committee of the Department of Psychology of LMU Munich approved the study.

Apparatus and stimuli

The experiment took place in a soundproof, dimly lit room (2.0 cd/m2) with a table lamp placed behind a 19‐inch CRT monitor (85 Hz refresh rate). The experiment was implemented using Matlab (Mathworks Inc.) and Psychtoolbox (Kleiner et al., 2007).

Duration stimuli were presented as Gabor patches displayed for a randomly selected duration from a uniform distribution of 500, 610, 740, 910, 1100, 1340, 1640 and 2000 ms, with equal spacing by 0.197 log units in the log‐scale. The geometric mean of the sampled durations was 1000 ms. The Gabor patches subtended 5° × 5° in visual angle and had a spatial frequency of 0.5 cycles per degree. Their orientation was randomly chosen from −81° to 81°, in 18° increments. The sinusoidal gratings of Gabor had a peak contrast of 100% Michelson and were enveloped by a Gaussian hull with a standard deviation of 2.29° radius. All stimuli and task cues appeared against a grey background (13.18 cd/m2).

Design and procedure

The experiment involved two tasks: duration reproduction and duration bisection. In the reproduction task, participants reproduced the duration of a presented stimulus as accurately as possible. In the bisection judgement task (Wearden, 1997), participants judged if the presented stimulus duration was shorter or longer than a standard duration (1 s), shown five times at the beginning of each experimental block.

Participants received task instructions only after the stimulus presentation following a post‐cue design: a dot indicated the reproduction task, while left‐ and right‐pointing arrows signalled the bisection task (see Figure 1). This design ensured consistent duration encoding across trials, independent of tasks, thus allowing us to assess if sequential effects generalized across task types.

FIGURE 1.

FIGURE 1

Open in a new tab

Schematic illustration of trial sequence. Each trial started with a 500‐ms central fixation, followed by a Gabor patch displayed for a given duration (500, 610, 740, 910, 1100, 1340, 1640, or 2000 ms). After a 300‐ms blank screen, a task cue appeared, informing the task for the current trial. In reproduction trials, observers pressed a key to reappear the same Gabor patch for the perceived duration. In bisection trials, they judged if the current duration was shorter or longer than the standard duration, shown multiple times at the start of each block. After the response, a random blank interval (500–1000 ms) was presented before the next trial.

The experiment comprised 20 blocks, each containing 32 trials, for a total of 640 trials. Each duration appeared 40 times per task type. Within each block, we randomized and balanced duration levels and task types, repeating each condition twice. To ensure equal representation, inter‐trial transitions from trial n−1 to trial n equally covered all possible task combinations: reproduction followed by reproduction (RR), bisection judgement followed by reproduction (JR), reproduction followed by bisection judgement (RJ), and judgement followed by judgement (JJ).

Each trial started with a 500‐ms central fixation, followed by a Gabor patch for a randomly selected duration. After a 300‐ms blank screen, a task cue, either a dot (reproduction task) or left‐ and right‐arrows (bisection task), appeared to indicate the task for the current trial. In the duration reproduction task, a dot cue remained visible until observers pressed a key. Upon keypress, the same Gabor patch reappeared, remaining on the screen until the key was released. The keypress duration indicated the reproduced duration. In the bisection judgement task, two directional arrows appeared, prompting observers to indicate via keypress whether the perceived duration was ‘shorter’ or ‘longer’ than the 1‐s standard. After each response, a blank inter‐trial interval (ITI) of 500–1000 ms preceded the next trial.

Prior to the formal experiment, observers practiced a block of 32 trials to familiarize themselves with the tasks. During practice, they received accuracy feedback lasting for 500 ms. For reproduction practice trials, feedback displayed ‘Too short’, ‘Well done’, or ‘Too long’ when the relative reproduction error ([Reproduced duration – Actual duration]/Actual duration) fell below −30%, within [−30%, 30%], or above 30%, respectively. In bisection trials, feedback indicated ‘Correct’ or ‘Incorrect’ based on whether the response matched the comparison between the presented duration and the 1‐s standard.

Statistical analysis

Considering that reproduced durations are influenced by both central tendency and serial dependence (approximately linear; Glasauer & Shi, 2022), we treated the reproduction bias (i.e. reproduced minus physical duration) as linearly driven by both the current and prior stimuli. Importantly, serial dependence combines two distinct processes: (a) sensory carryover, a bias from the prior stimulus; and (b) decisional carryover, an attractive bias towards the prior response. However, the direction of sensory carryover is rather mixed. Some found repulsive biases (Fritsche et al., 2017; Pascucci et al., 2019; Sadil et al., 2024), while others showed attractive biases (Cheng, Chen, & Shi, 2024; Cheng, Chen, Yang, & Shi, 2024; Fischer & Whitney, 2014; Glasauer & Shi, 2022). These mixed findings could be a methodological artefact: when both prior stimulus and prior response are considered, while sharing covariance due to their inherent correlation, potentially leading to multicollinearity. Consequently, the true stimulus‐based sensory carryover of serial dependence may be masked or cancelled out by decisional carryover of prior responses with standard regressions.

To this end, we applied structural equation modelling (SEM), which offers several key advantages over standard regression models. It allows explicit modelling of latent structures, enabling causal modelling via directed paths (see Figure 2): current stimulus → current response; prior stimulus → bias (direct sensory carryover), and prior report → current response (decisional carryover). This model also includes an indirect sensory carryover: prior stimulus → prior report → current response, mirroring theoretical processes of sensory versus decisional serial dependencies. Unlike conventional regression, SEM incorporates measurement error, enhancing parameter reliability. By explicitly modelling the covariance between the prior stimulus and the prior response, SEM alleviates multicollinearity issues. To our knowledge, no prior study has used SEM to dissect sensory versus decisional carryover in this manner.

FIGURE 2.

FIGURE 2

Open in a new tab

The schematic of structural equation modelling (SEM). The model specifies directed paths: Current stimulus → current response (𝛽1); prior stimulus → bias (direct sensory carryover, 𝛽2); and prior report → current response (decisional carryover, 𝛽3). This model also includes an indirect sensory carryover: Prior stimulus → prior report → current response (α 1*𝛽3), mirroring theoretical processes of sensory versus decisional serial dependencies. The causal influences can be traced to earlier trials (e.g. trials n‐2).

Specifically, we constructed structural paths to parse the causal influences of current duration (𝛽 1), prior duration (𝛽 2), and prior report (short/long category) in trial n−1 (𝛽 3) on the current response (Figure 2). Such causal influences can be traced back to earlier trials. To unify the two task types, we assumed that decisional carryover mainly stems from the coarse categorization of the prior report (short or long). In the bisection task, this categorization was explicit; in the reproduction task, we treated it as implicit, based on how responses compared to their mean reproduction. To account for the scalar property in duration estimation (Gibbon et al., 1984; Ren et al., 2021), we modelled time duration on a log‐scale.

We applied SEM to both tasks using the lavaan package in R for model specification, estimation and evaluation (Rosseel, 2012). The categorical variables (e.g. the prior report and the bisection response) were treated as ordinal endogenous variables. Lavaan automatically applies a threshold SEM using the diagonally weighted least squares with mean–variance adjustment (WLSMV) estimator. For binary responses in the bisection task, this is formally equivalent to a probit regression. That is, lavaan models an underlying continuous latent variable Y* (the latent response) with thresholds, and the probability of a positive (‘long’) response is PY*>0X=ΦXTβ, where Φ is the standard normal cumulative distribution function. This model is analogous to the conventional measure estimating sequential effect in binary judgement tasks—the probabilistic choice model (Feigin, Baror, et al., 2021; Feigin, Shalom‐Sperber, et al., 2021; Li et al., 2023), which takes into account current stimuli, prior stimuli and prior report together (but potentially encountering multicollinearity issues). To compare with the conventional probabilistic choice model, we also conducted the conventional approach (see Appendix S1).

RESULTS

The missed trials in the bisection task and trials with extreme reproduced durations (i.e. shorter than 50 ms or exceeding 5 s) were excluded from the analysis (0.40% in total). As for the four task combinations for trial n−1 and n (i.e. RR, JR, RJ and JJ), the inter‐trial transitional probability was equal but with a slight cross‐subject variance due to randomization (25.00% ± 1.17%).

Reproduction task

We applied SEMs to individual participants on each prior task type (reproduction vs. bisection). The model provided excellent fits to the data, with mean χ 2(3) values of 3.07 for the RR condition and 3.20 for the JR condition (ps > 0.508). The mean Goodness of Fit Indices (GFIs) were 0.988 and 0.989 for the RR and JR, respectively. These indices meet conventional thresholds, supporting acceptable model performance. Table 1 presents the mean coefficients from the SEM models, along with related statistical results (see also Figure 3).

TABLE 1.

Estimates of the structural equation modelling (SEM) for the reproduction task.

RR (M ± SE) JR (M ± SE) t df p Cohen's d BF10
𝛽1 −0.421 ± 0.050*** −0.409 ± 0.050*** −0.751 23 .460 −0.153 0.277
𝛽2 −0.212 ± 0.037*** −0.024 ± 0.041 −4.357 23 < .001 −0.889 > 100
𝛽3 0.097 ± 0.018*** 0.010 ± 0.012 5.096 23 < .001 1.040 > 100
𝛽4 0.251 ± 0.034*** 0.025 ± 0.039 5.181 23 < .001 1.058 > 100
𝛽5 0.038 ± 0.013** 0.001 ± 0.013 3.047 23 .006 0.622 7.765
Open in a new tab

Note: The parameter 𝛽4 represents the indirect mediating effect (𝛽4 = α 1 * 𝛽3) and 𝛽5 represents the net serial dependence (net effect from prior duration: 𝛽5 = 𝛽2 + α 1 * 𝛽3). The statistics, including t, p, df, Cohen's d and Bayes Factor, are based on the comparison between the relevance conditions of prior task type (i.e. RR vs. JR). Stars ** and *** represent p < .01 and .001, respectively, for statistical significance compared to zero.

FIGURE 3.

FIGURE 3

Open in a new tab

Reproduction task performance. (a) Estimated coefficients in SEM for the reproduction task. The first row shows the predictors from a prior trial, and the second shows predictors in the current trial. The solid lines and coefficients in dark red show the parameter estimates, which describe the strength and direction of the relationships between variables, for the task‐relevant (RR) trials. The dashed lines and coefficients in orange depict the estimates for the task‐irrelevant (JR) trials. (b) Parameter estimates for prior task types, predicted by the structural equation modelling (SEM). The error bar shows standard errors across participants. ***p < .001, **p < .01.

Direct effects

The central tendency effect (𝛽1): The current log‐transformed duration showed a significant negative linear relationship with reproduction bias in both RR (−0.421) and JR (−0.409) conditions (Table 1), confirming the classic central tendency biases. However, the central tendency biases were comparable between the two conditions (p = .460).

Perceptual serial dependence (𝛽2): While the prior duration had a significantly negative slope (−0.212) on the current bias in the RR task‐congruent sequence (p < .001), it did not show significance in the JR task‐incongruent sequence (p = .567). The negative slope shows that longer prior durations caused a negative bias, reflecting a common repulsion effect. The difference between the two conditions was also significant (p < .001, Figure 3b), suggesting that prior task relevance is an important factor in perceptual serial dependence.

Decisional carryover (𝛽3): Similar to perceptual serial dependence, decisional carryover also exhibited differential effects in two conditions. But instead of negative repulsion, the carryover was a significantly positive attraction (0.097) in the RR condition (p < .001), but non‐significant in the JR condition (p = .408). The difference between the two conditions was also significant (p < .001). The pattern suggests that decisional carryover occurs only with relevant tasks.

The indirect effect

One advantage of using SEM is that it allows us to estimate the indirect effect. A significant indirect effect emerged via the pathway from prior duration → prior report → reproduction bias (𝛽4 = α 1 * β 3) in the RR condition (p < .001) but failed to show any effect in the JR condition (p = .540). The difference between the two conditions was also significant (p < .001).

The net serial dependence

The net effect from prior duration (𝛽5 = 𝛽2 + α 1 * 𝛽3) was consistent with the serial dependence measured by conventional regression (see Appendix S1). The net serial dependence showed significance in the RR (p = .006) but not the JR (p = .943) condition. The difference between the task relevance conditions was also significant (p = .006). This further suggests that conventional serial dependence could be the cancellation of perceptual serial dependence and decisional carryover.

Bisection judgement task

The generalized SEMs (probit models) for the bisection task also provided excellent fits to the data, with mean χ 2(3) values of 2.62 for the RJ condition and 1.93 for the JJ condition (ps > 0.542). The mean Goodness of Fit Indices (GFIs) were 0.960 and 0.975 for the RJ and JJ conditions, respectively. Table 2 presents the mean coefficients from the generalized SEM models, along with related statistical results (see also Figure 4).

TABLE 2.

Estimates of the structural equation modelling (SEM) for the judgement task.

JJ (M ± SE) RJ (M ± SE) t df p Cohen's d BF 10
𝛽1 3.427 ± 0.285*** 3.227 ± 0.279*** 1.330 23 .196 0.272 0.469
𝛽2 −0.882 ± 0.233*** 0.030 ± 0.194 −2.482 23 .021 −0.507 2.634
𝛽3 0.258 ± 0.071** 0.041 ± 0.060 2.183 23 .039 0.446 1.568
𝛽4 0.884 ± 0.243** 0.036 ± 0.193 2.384 23 .026 0.487 2.214
𝛽5 0.002 ± 0.073 0.065 ± 0.108 −0.694 23 .495 −0.142 0.267
Open in a new tab

Note: The parameter 𝛽4 represents the indirect mediating effect (𝛽4 = α 1 * 𝛽3) and 𝛽5 represents the net serial dependence (net effect from prior duration: 𝛽5 = 𝛽2 + α 1 * 𝛽3). The statistics—t, p, df, Cohen's d and Bayes Factor—compare the relevance conditions of the prior task type (RR vs. JR). The symbols ** and *** indicate significance levels of p < .01 and .001, respectively, compared to zero.

FIGURE 4.

FIGURE 4

Open in a new tab

Judgement task performance. (a) Estimated coefficients in SEM for the judgement task. The first row shows the factors from a prior trial, and the second shows factors in the current trial. The task‐relevant coefficients (JJ) are with solid lines and in dark red, while the task‐irrelevant coefficients (RJ) are with dashed lines and in orange. (b) Parameter estimates for prior task types predicted by the structural equation modelling (SEM). The error bar shows the standard error across participants. *p < .05.

It is important to note that the coefficients 𝛽1, 𝛽2 and 𝛽3 correspond conceptually to central tendency, perceptual serial dependence, and decisional carryover, respectively; they do not directly measure those effects in the observed data scale. Rather, they quantify the impact of predictors on the latent response variable Y* (a z‐score scale), not on the raw behavioural bias. Specifically, 𝛽1 expresses how a one‐unit increase in log‐transformed current duration shifts the latent decision propensity Y* towards ‘long’ by 𝛽1 standard deviations. Table 2 shows that both RJ and JJ had comparable 𝛽1 (p = .196).

The coefficients 𝛽2 and 𝛽3 similarly reflect how prior stimulus and prior response influence that latent decision tendency. Similar to the reproduction task, the prior duration had a significantly negative repulsion on the latent response variable Y* in the JJ task‐congruent sequence (p < .001); it did not show significance in the RJ task‐incongruent sequence (p = .882). The difference between the two conditions was also significant (p = .021, Figure 4b). The decisional carryover on the latent variable also exhibited differential effects in two conditions. But instead of negative repulsion, the carryover was significantly positive (i.e., attractive) in the JJ condition (p = .001), but not in the JR condition (p = .506). The difference between the two conditions was also significant (p = .039). The pattern suggests that decisional carryover in the latent space occurs only with relevant tasks.

The coefficient β 4, reflected the indirect effect via the pathway from prior duration → prior report → reproduction bias, showed significance in the JJ (p = .001) but not RJ (p = .855) sequence. The comparison between task‐congruence conditions also reached significance (p = .026). Intriguingly, the coefficient β 5 which reflects the net sequential effect on latent decision tendency showed no effects in both conditions (ps > .555) and the between‐condition comparison (p = .495). These net effects on latent decision are in line with standard behavioural observations of serial dependence (i.e. the point of subjective equality, see Appendix S1).

DISCUSSION

Previous research has yielded contradictory findings about whether motor responses are required for serial dependence: Fischer and Whitney (2014) found serial dependence persisted without responses, while Pascucci et al. (2019) found attractive biases disappeared when responses were withheld. To resolve this controversy, we tested whether serial dependence emerges when tasks share identical stimulus features but differ only in response modes. We interleaved duration reproduction and bisection tasks, employing a post‐cue design to ensure identical perceptual encoding while varying motor output across trials. At least for temporal judgements, response‐mode consistency—not shared memory encoding alone—drives sequential biases.

Within each task, we observed strong sequential effects. In the reproduction task, participants exhibited a repulsion effect in perceptual serial dependence and an attraction effect in decisional carryover. Similarly, in the bisection task, we observed a repulsion in latent decision propensity and positive decisional carryover, consistent with previous findings (Li et al., 2023; Wiener et al., 2014). Critically, both effects vanished when tasks alternated, despite identical stimulus processing. These findings support our BRAC‐based hypothesis: response‐feature binding drives serial dependence, and switching response modes disrupts this binding, eliminating cross‐task sequential biases.

The perceptual repulsive bias observed in our within‐task condition replicates findings from previous studies (Moon & Kwon, 2022; Sadil et al., 2024). Such repulsive bias is typically attributed to low‐level sensory processing, such as neural adaptation (Moon & Kwon, 2022) or efficient encoding mechanisms that amplify differences between stimulus representations (Wei & Stocker, 2015). However, many studies also demonstrate attractive serial dependence (Cheng, Chen, Glasauer, et al., 2024; Cheng, Chen, & Shi, 2024; Cheng, Chen, Yang, & Shi, 2024; Glasauer & Shi, 2022; Wang et al., 2023), including the seminal study (Fischer & Whitney, 2014). The co‐occurrence of repulsive and attractive effects complicates interpretation when using traditional one‐factor models, which typically estimate net serial dependence—a mixture of opposing biases from different processing stages (Fritsche et al., 2020; Moon & Kwon, 2022).

By contrast, Structural Equation Modelling (SEM) enables us to separate sensory‐level effects (direct repulsion from prior stimulus) from decision‐level effects (attraction towards prior response), as portrayed in Figure 2. Specifically, our SEM framework models both direct pathways from prior stimulus to current report (capturing perceptual repulsion) and indirect pathways that mediate via prior response (capturing decision carryover). When summing these paths, we found that within‐task net serial dependence becomes attractive in reproduction tasks (0.038 ± 0.013 in RR reproduction). This mirrors classic one‐factor estimates but importantly hides the underlying sensory repulsion. Indeed, Sadil et al. (2024) cautioned that one‐factor models can be misleading because they collapse distinct mechanisms into a single summary metric.

Even conventional regression analyses that include both prior stimulus and prior response often suffer from multicollinearity (as two factors are highly correlated), making it difficult to separately estimate repulsive and attractive contributions. In our work, we also conducted such regression analyses (see Appendix S1) and observed results largely aligned with but also diverged from SEM outcomes. For instance, conventional regression sometimes suggested ‘attractive’ perceptual serial dependence in reproduction—likely because it cannot model the indirect path via decision carryover, and the positive prior response effect masks repulsion. The capacity of SEM to model latent constructs, measurement error and causal pathways allows it to disentangle repulsive sensory effects that conventional methods miss.

The repulsive effect we observed within the bisection task also replicates earlier findings (Li et al., 2023). However, it is crucial to note that this ‘repulsion’ occurs at the level of the latent decision propensity, not directly in the observed categorical responses. In other words, while prior durations shift the underlying probability of responding ‘long’ away from the previous stimulus, this does not necessarily translate into final choice‐level repulsion. Indeed, in our data the point of subjective equality (PSE), which indexes the actual bisection threshold, remained unchanged across prior task conditions (see Appendix S1). Therefore, caution is warranted: our probabilistic choice model reveals repulsion in latent decision tendencies, but this should not be equated with behavioural repulsion in overt judgements. The repulsive effect at the latent level may influence underlying discrimination sensitivity without affecting the observable bisection performance.

The key finding of the current study is the response‐mode‐specific serial dependence. At first glance, this may resemble previous findings that show no sequential effects across tasks (e.g. across numerosity and duration tasks in Fornaciai et al., 2023; Togoli et al., 2021) or across modalities (Li et al., 2023). However, duration representation likely differs from the representation of the other dimension (e.g. numerosity) or is encoded in a modality‐specific manner at the low‐level physiological processing. Since both task‐related features (e.g. duration and numerosity) can be maintained simultaneously in working memory, previous reports of task‐specific serial dependence may be attributed to an early stage dimension‐ or modality‐specific processing pathway (Togoli et al., 2021), occurring before the integration of sensory likelihood and prior knowledge in memory (Körding & Wolpert, 2004; Weiss et al., 2002). However, unlike these previous studies, the present study employed the same duration within the same visual modality for both timing tasks, ruling out this interpretation. Instead, our findings highlight response‐specific serial dependence, suggesting that a separate, high‐level processing stage occurs after early stage duration encoding. The mechanisms underlying different consecutive task combinations (i.e. RR vs. JR and JJ vs. RJ) may probably differ in high‐level processing after duration encoding, likely due to the reactivation of response‐related tasks. Considering that both tasks exhibit attractive decisional carryover, it is more plausible that observers engage distinct response‐dependent high‐level processing for duration reproduction and bisection, making sequential effects from cross‐task transitions less influential.

Previous research suggests that processing current inputs can reactivate memory of prior experiences (Bae & Luck, 2019; St. John‐Saaltink et al., 2016), which may then integrate with ongoing memory representations (Zhang & Lewis‐Peacock, 2024). Importantly, the strength of this memory reactivation has been linked to the strength of sequential bias (Barbosa & Compte, 2020). It is therefore possible that retrieval of task‐related responses only reactivates prior duration with the same response mode in working memory. This is consistent with the theory of binding and retrieval in action control (BRAC, Frings et al., 2020), which suggests that stimulus features and responses are bound together as an event file and retrieving the same response will activate the whole event file (Allenmark et al., 2025; Hommel, 2004). This could also explain previous findings of multiple motor‐specific priors in interleaved duration reproduction and bisection tasks (Roach et al., 2017). Neuroimaging evidence further supports this idea. Studies have shown topographically organized duration maps in the human supplementary motor area (SMA, Kulashekhar et al., 2021; Protopapa et al., 2019), tuned to temporal contexts. The SMA is crucial for motor preparation and time perception (Coull et al., 2011; Kotz & Schwartze, 2011; Merchant et al., 2013). Additionally, sequential bias has been linked to activities in the left middle frontal gyrus (Cheng, Chen, Glasauer, et al., 2024), a region associated with the fronto‐striatal pathway critical for working memory and time perception (Matell et al., 2005). These behavioural and neural findings imply the possibility of response‐related memory reactivations in duration reproduction and bisection tasks.

The current findings do not contradict previous interpretations of cross‐task serial dependence. Rather, they clarify the locus at which prior experience interacts with current processing: responses engage with prior experiences at a later stage, after encoding. If prior history biased duration encoding directly, we would anticipate significant cross‐task sequential effects, given that both tasks shared identical encoding. Instead, our results reveal differential effects across task transitions, pointing to a response‐specific processing stage that mediates serial dependence after early encoding. This response binding was also evident in the decisional carryover we observed: making a ‘long’ or ‘short’ response in the previous trial, in both reproduction and bisection, attracted the current response in that direction, replicating many previous findings (Cheng, Chen, Yang, & Shi, 2024; Li et al., 2023; Wehrman et al., 2023; Wehrman et al., 2020; Wiener et al., 2014).

In orientation perception, serial dependence transfers across task types (Fischer & Whitney, 2014; Fornaciai & Park, 2018; Liberman et al., 2016; St. John‐Saaltink et al., 2016), persists in one‐trial paradigms without prior responses (Manassi & Whitney, 2022), emerges in detection tasks where the feature is not task‐relevant (Murai & Whitney, 2021) and exhibits spatial tuning that decreases with distance between stimuli (Fischer & Whitney, 2014; Pascucci et al., 2023), consistent with perceptual rather than decisional origins. Cicchini et al. (2021) attributed such cross‐context effects to a hierarchical predictive coding framework, in which context‐biased expectations feed back to lower levels of sensory processing (Friston, 2009; Rao & Ballard, 1999; Summerfield & de Lange, 2014). These findings collectively establish that an early perceptual component of serial dependence can operate independently of response requirements, at least in domains with relatively autonomous sensory processing. Temporal processing, however, differs. Duration perception relies more heavily on neural circuits that overlap with motor planning—including the supplementary motor area and fronto‐striatal pathways (Coull et al., 2011; Merchant et al., 2013)—than orientation does. This tighter coupling between temporal representation and motor processes may explain why even the perceptual component of serial dependence vanished across tasks in our data: under BRAC, the entire event file—stimulus features and response mode—is bound together (Frings et al., 2020), and switching response modes disrupts retrieval of the complete representation. In domains like orientation, where sensory processing is more autonomous from motor planning, the perceptual component may survive task switches, producing the cross‐task effects documented in the literature. The balance between perceptual and decisional contributions to serial dependence thus likely varies across feature domains, shaped by the degree of sensory–motor coupling.

In summary, our systematic task‐switching design reveals that serial dependence requires response‐mode consistency: sequential biases emerged only when motor responses were repeated, even though stimulus features were identical across tasks. This pattern supports the BRAC framework, demonstrating that stimulus features and responses are bound together as event files in working memory, and that reactivation of past experiences depends critically on retrieving the same response mode. In the timing domain, where stimulus representation and motor processing are tightly coupled, response‐feature binding is the dominant mechanism driving sequential biases. Whether the same holds in domains with more autonomous sensory processing remains an open and important question.

AUTHOR CONTRIBUTIONS

Jiao Wu: Conceptualization; methodology; software; data curation; formal analysis; visualization; project administration; writing – original draft; writing – review and editing; investigation. Halid Oğuz Serçe: Investigation; visualization. Zhuanghua Shi: Conceptualization; methodology; formal analysis; supervision; funding acquisition; resources; writing – review and editing.

Supporting information

Figure S1.

BJOP-117-1396-s001.docx (583.6KB, docx)

ACKNOWLEDGEMENTS

This study was supported by the German Science Foundation (DFG) research grants Sh166/10‐1 to Z.S., the China Scholarship Council (CSC) Scholarship to J.W., and the Erasmus+ internship to H.O.S. Open Access funding enabled and organized by Projekt DEAL.

DATA AVAILABILITY STATEMENT

The data that support the findings of this study are openly available at https://github.com/Jiao2048/DurSeqEff_RprBscTasks.

REFERENCES

  1. Allenmark, F. , Yu, H. , Müller, H. J. , Shi, Z. , & Frings, C. (2025). Distractor‐response binding influences visual search. Attention, Perception & Psychophysics, 56(4), 1047–1070. 10.3758/s13414-025-03016-7 [DOI] [PubMed] [Google Scholar]
  2. Bae, G.‐Y. , & Luck, S. J. (2019). Reactivation of previous experiences in a working memory task. Psychological Science, 30(4), 587–595. 10.1177/0956797619830398 [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Bae, G.‐Y. , & Luck, S. J. (2020). Serial dependence in vision: Merely encoding the previous‐trial target is not enough. Psychonomic Bulletin & Review, 27(2), 293–300. 10.3758/s13423-019-01678-7 [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Barbosa, J. , & Compte, A. (2020). Build‐up of serial dependence in color working memory. Scientific Reports, 10(1), 10959. 10.1038/s41598-020-67861-2 [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Bertolasi, J. , Esposito, D. , Vitale, A. , & Gori, M. (2025). Exploring the role of serial dependence in visual time perception. Journal of Vision, 25(8), 7. 10.1167/jov.25.8.7 [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Bliss, D. P. , Sun, J. J. , & D'Esposito, M. (2017). Serial dependence is absent at the time of perception but increases in visual working memory. Scientific Reports, 7(1), 14739. 10.1038/s41598-017-15199-7 [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Burr, D. , & Cicchini, G. M. (2014). Vision: Efficient adaptive coding. Current Biology: CB, 24(22), R1096–R1098. 10.1016/j.cub.2014.10.002 [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Ceylan, G. , & Pascucci, D. (2023). Attractive and repulsive serial dependence: The role of task relevance, the passage of time, and the number of stimuli. Journal of Vision, 23(6), 8. 10.1167/jov.23.6.8 [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Chen, S. , Wang, T. , & Bao, Y. (2023). Serial dependence in timing at the perceptual level being modulated by working memory. PsyCh Journal, 12(6), 774–786. 10.1002/pchj.653 [DOI] [PubMed] [Google Scholar]
  10. Cheng, S. , Chen, S. , Glasauer, S. , Keeser, D. , & Shi, Z. (2024). Neural mechanisms of sequential dependence in time perception: The impact of prior task and memory processing. Cerebral Cortex, 34(1), bhad453. 10.1093/cercor/bhad453 [DOI] [PubMed] [Google Scholar]
  11. Cheng, S. , Chen, S. , & Shi, Z. (2024). Opposing sequential biases in direction and time reproduction: Influences of task relevance and working memory. British Journal of Psychology, 115(4), 825–842. 10.1111/bjop.12728 [DOI] [PubMed] [Google Scholar]
  12. Cheng, S. , Chen, S. , Yang, X. , & Shi, Z. (2024). The impact of task measurements on sequential dependence: A comparison between temporal reproduction and discrimination tasks. Psychological Research Psychologische Forschung, 88, 2346–2359. 10.1007/s00426-024-02023-x [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Cicchini, G. M. , Benedetto, A. , & Burr, D. C. (2021). Perceptual history propagates down to early levels of sensory analysis. Current Biology: CB, 31(6), 1245–1250. 10.1016/j.cub.2020.12.004 [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Cicchini, G. M. , Mikellidou, K. , & Burr, D. (2017). Serial dependencies act directly on perception. Journal of Vision, 17(14), 6. 10.1167/17.14.6 [DOI] [PubMed] [Google Scholar]
  15. Cicchini, G. M. , Mikellidou, K. , & Burr, D. C. (2024). Serial dependence in perception. Annual Review of Psychology, 75(1), 129–154. 10.1146/annurev-psych-021523-104939 [DOI] [PubMed] [Google Scholar]
  16. Coull, J. T. , Cheng, R.‐K. , & Meck, W. H. (2011). Neuroanatomical and neurochemical substrates of timing. Neuropsychopharmacology, 36(1), 3–25. 10.1038/npp.2010.113 [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Erdfelder, E. , Faul, F. , & Buchner, A. (1996). GPOWER: A general power analysis program. Behavior Research Methods, Instruments, & Computers: A Journal of the Psychonomic Society, Inc, 28(1), 1–11. 10.3758/bf03203630 [DOI] [Google Scholar]
  18. Feigin, H. , Baror, S. , Bar, M. , & Zaidel, A. (2021). Perceptual decisions are biased toward relevant prior choices. Scientific Reports, 11(1), 648. 10.1038/s41598-020-80128-0 [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Feigin, H. , Shalom‐Sperber, S. , Zachor, D. A. , & Zaidel, A. (2021). Increased influence of prior choices on perceptual decisions in autism. eLife, 10, e61595. 10.7554/eLife.61595 [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Fischer, C. , Kaiser, J. , & Bledowski, C. (2024). A direct neural signature of serial dependence in working memory. eLife, 13, RP99478. 10.7554/elife.99478.1 [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Fischer, J. , & Whitney, D. (2014). Serial dependence in visual perception. Nature Neuroscience, 17(5), 738–743. 10.1038/nn.3689 [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Fornaciai, M. , & Park, J. (2018). Attractive serial dependence in the absence of an explicit task. Psychological Science, 29(3), 437–446. 10.1177/0956797617737385 [DOI] [PubMed] [Google Scholar]
  23. Fornaciai, M. , Togoli, I. , & Bueti, D. (2023). Perceptual history biases are predicted by early visual‐evoked activity. The Journal of Neuroscience: The Official Journal of the Society for Neuroscience, 43(21), 3860–3875. [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Frings, C. , Hommel, B. , Koch, I. , Rothermund, K. , Dignath, D. , Giesen, C. , Kiesel, A. , Kunde, W. , Mayr, S. , Moeller, B. , Möller, M. , Pfister, R. , & Philipp, A. (2020). Binding and retrieval in action control (BRAC). Trends in Cognitive Sciences, 24(5), 375–387. 10.1016/j.tics.2020.02.004 [DOI] [PubMed] [Google Scholar]
  25. Friston, K. (2009). The free‐energy principle: A rough guide to the brain? Trends in Cognitive Sciences, 13(7), 293–301. 10.1016/j.tics.2009.04.005 [DOI] [PubMed] [Google Scholar]
  26. Fritsche, M. , Mostert, P. , & de Lange, F. P. (2017). Opposite effects of recent history on perception and decision. Current Biology: CB, 27(4), 590–595. 10.1016/j.cub.2017.01.006 [DOI] [PubMed] [Google Scholar]
  27. Fritsche, M. , Spaak, E. , & de Lange, F. P. (2020). A Bayesian and efficient observer model explains concurrent attractive and repulsive history biases in visual perception. eLife, 9, e55389.  10.7554/eLife.55389 [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Gallagher, G. K. , & Benton, C. P. (2024). Characterizing serial dependence as an attraction to prior response. Journal of Vision, 24(9), 16. 10.1167/jov.24.9.16 [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Gibbon, J. , Church, R. M. , & Meck, W. H. (1984). Scalar timing in memory. Annals of the New York Academy of Sciences, 423, 52–77. 10.1111/j.1749-6632.1984.tb23417.x [DOI] [PubMed] [Google Scholar]
  30. Glasauer, S. , & Shi, Z. (2022). Individual beliefs about temporal continuity explain variation of perceptual biases. Scientific Reports, 12(1), 10746. 10.1038/s41598-022-14939-8 [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Hahn, M. , & Wei, X.‐X. (2024). A unifying theory explains seemingly contradictory biases in perceptual estimation. Nature Neuroscience, 27(4), 793–804. 10.1038/s41593-024-01574-x [DOI] [PubMed] [Google Scholar]
  32. Hajonides, J. E. , van Ede, F. , Stokes, M. G. , Nobre, A. C. , & Myers, N. E. (2023). Multiple and dissociable effects of sensory history on working‐memory performance. The Journal of Neuroscience: The Official Journal of the Society for Neuroscience, 43(15), 2730–2740. 10.1523/JNEUROSCI.1200-22.2023 [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Hommel, B. (2004). Event files: Feature binding in and across perception and action. Trends in Cognitive Sciences, 8(11), 494–500. 10.1016/j.tics.2004.08.007 [DOI] [PubMed] [Google Scholar]
  34. Ivry, R. B. , & Schlerf, J. E. (2008). Dedicated and intrinsic models of time perception. Trends in Cognitive Sciences, 12(7), 273–280. 10.1016/j.tics.2008.04.002 [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Kiyonaga, A. , Scimeca, J. M. , Bliss, D. P. , & Whitney, D. (2017). Serial dependence across perception, attention, and memory. Trends in Cognitive Sciences, 21(7), 493–497. 10.1016/j.tics.2017.04.011 [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. Kleiner, M. , Brainard, D. , & Pelli, D. (2007). What's new in Psychtoolbox‐3? Perception ECVP Abstract Supplement, 36, 14. 10.1177/03010066070360S101 [DOI] [Google Scholar]
  37. Körding, K. P. , & Wolpert, D. M. (2004). Bayesian integration in sensorimotor learning. Nature, 427(6971), 244–247. 10.1038/nature02169 [DOI] [PubMed] [Google Scholar]
  38. Kotz, S. A. E. , & Schwartze, M. (2011). Differential input of the supplementary motor area to a dedicated temporal processing network: Functional and clinical implications. Frontiers in Integrative Neuroscience, 5, 86. 10.3389/fnint.2011.00086 [DOI] [PMC free article] [PubMed] [Google Scholar]
  39. Kulashekhar, S. , Maass, S. , Van Rijn, H. , & Bueti, D. (2021). The topographic representation of time and its link with temporal context and perception. In Research Square. Research Square Company. 10.21203/rs.3.rs-753370/v1 [DOI] [Google Scholar]
  40. Li, B. , Wang, B. , & Zaidel, A. (2023). Modality‐specific sensory and decisional carryover effects in duration perception. BMC Biology, 21(1), 48. 10.1186/s12915-023-01547-9 [DOI] [PMC free article] [PubMed] [Google Scholar]
  41. Liberman, A. , Zhang, K. , & Whitney, D. (2016). Serial dependence promotes object stability during occlusion. Journal of Vision, 16(15), 16. 10.1167/16.15.16 [DOI] [PMC free article] [PubMed] [Google Scholar]
  42. Luo, M. , Zhang, H. , Fang, F. , & Luo, H. (2025). Reactivation of previous decisions repulsively biases sensory encoding but attractively biases decision‐making. PLoS Biology, 23(4), e3003150. 10.1371/journal.pbio.3003150 [DOI] [PMC free article] [PubMed] [Google Scholar]
  43. Manassi, M. , Liberman, A. , Kosovicheva, A. , Zhang, K. , & Whitney, D. (2018). Serial dependence in position occurs at the time of perception. Psychonomic Bulletin & Review, 25(6), 2245–2253. 10.3758/s13423-018-1454-5 [DOI] [PubMed] [Google Scholar]
  44. Manassi, M. , Murai, Y. , & Whitney, D. (2023). Serial dependence in visual perception: A meta‐analysis and review. Journal of Vision, 23(8), 18. 10.1167/jov.23.8.18 [DOI] [PMC free article] [PubMed] [Google Scholar]
  45. Manassi, M. , & Whitney, D. (2022). Illusion of visual stability through active perceptual serial dependence. Science Advances, 8(2), eabk2480. 10.1126/sciadv.abk2480 [DOI] [PMC free article] [PubMed] [Google Scholar]
  46. Markov, Y. A. , Tiurina, N. A. , & Pascucci, D. (2024). Serial dependence: A matter of memory load. Heliyon, 10(13), e33977. 10.1016/j.heliyon.2024.e33977 [DOI] [PMC free article] [PubMed] [Google Scholar]
  47. Matell, M. S. , Meck, W. H. , & Lustig, C. (2005). Not “just” a coincidence: Frontal‐striatal interactions in working memory and interval timing. Memory, 13(3‐4), 441–448. 10.1080/09658210344000404 [DOI] [PubMed] [Google Scholar]
  48. Merchant, H. , Harrington, D. L. , & Meck, W. H. (2013). Neural basis of the perception and estimation of time. Annual Review of Neuroscience, 36(1), 313–336. 10.1146/annurev-neuro-062012-170349 [DOI] [PubMed] [Google Scholar]
  49. Moon, J. , & Kwon, O.‐S. (2022). Attractive and repulsive effects of sensory history concurrently shape visual perception. BMC Biology, 20(1), 247. 10.1186/s12915-022-01444-7 [DOI] [PMC free article] [PubMed] [Google Scholar]
  50. Murai, Y. , & Whitney, D. (2021). Serial dependence revealed in history‐dependent perceptual templates. Current Biology: CB, 31(14), 3185–3191. 10.1016/j.cub.2021.05.006 [DOI] [PMC free article] [PubMed] [Google Scholar]
  51. Pascucci, D. , Mancuso, G. , Santandrea, E. , Della Libera, C. , Plomp, G. , & Chelazzi, L. (2019). Laws of concatenated perception: Vision goes for novelty, decisions for perseverance. PLoS Biology, 17(3), e3000144. 10.1371/journal.pbio.3000144 [DOI] [PMC free article] [PubMed] [Google Scholar]
  52. Pascucci, D. , & Plomp, G. (2021). Author correction: Serial dependence and representational momentum in single‐trial perceptual decisions. Scientific Reports, 11(1), 16750. 10.1038/s41598-021-96023-1 [DOI] [PMC free article] [PubMed] [Google Scholar]
  53. Pascucci, D. , Tanrikulu, Ö. D. , Ozkirli, A. , Houborg, C. , Ceylan, G. , Zerr, P. , Rafiei, M. , & Kristjánsson, Á. (2023). Serial dependence in visual perception: A review. Journal of Vision, 23(1), 9. 10.1167/jov.23.1.9 [DOI] [PMC free article] [PubMed] [Google Scholar]
  54. Protopapa, F. , Hayashi, M. J. , Kulashekhar, S. , van der Zwaag, W. , Battistella, G. , Murray, M. M. , Kanai, R. , & Bueti, D. (2019). Chronotopic maps in human supplementary motor area. PLoS Biology, 17(3), e3000026. 10.1371/journal.pbio.3000026 [DOI] [PMC free article] [PubMed] [Google Scholar]
  55. Rao, R. P. , & Ballard, D. H. (1999). Predictive coding in the visual cortex: A functional interpretation of some extra‐classical receptive‐field effects. Nature Neuroscience, 2(1), 79–87. 10.1038/4580 [DOI] [PubMed] [Google Scholar]
  56. Ren, Y. , Allenmark, F. , Müller, H. J. , & Shi, Z. (2021). Variation in the “coefficient of variation”: Rethinking the violation of the scalar property in time‐duration judgments. Acta Psychologica, 214, 103263. 10.1016/j.actpsy.2021.103263 [DOI] [PubMed] [Google Scholar]
  57. Roach, N. W. , McGraw, P. V. , Whitaker, D. J. , & Heron, J. (2017). Generalization of prior information for rapid Bayesian time estimation. Proceedings of the National Academy of Sciences, 114(2), 412–417. 10.1073/pnas.1610706114 [DOI] [PMC free article] [PubMed] [Google Scholar]
  58. Rosseel, Y. (2012). Lavaan: An R Package for structural equation modeling. Journal of Statistical Software, 48(2), 1–36. 10.18637/jss.v048.i02 [DOI] [Google Scholar]
  59. Sadil, P. , Cowell, R. A. , & Huber, D. E. (2024). The push‐pull of serial dependence effects: Attraction to the prior response and repulsion from the prior stimulus. Psychonomic Bulletin & Review, 31(1), 259–273. 10.3758/s13423-023-02320-3 [DOI] [PMC free article] [PubMed] [Google Scholar]
  60. Sheehan, T. C. , & Serences, J. T. (2022). Attractive serial dependence overcomes repulsive neuronal adaptation. PLoS Biology, 20(9), e3001711. 10.1371/journal.pbio.3001711 [DOI] [PMC free article] [PubMed] [Google Scholar]
  61. St. John‐Saaltink, E. , Kok, P. , Lau, H. C. , & de Lange, F. P. (2016). Serial dependence in perceptual decisions is reflected in activity patterns in primary visual cortex. The Journal of Neuroscience: The Official Journal of the Society for Neuroscience, 36(23), 6186–6192. 10.1523/jneurosci.4390-15.2016 [DOI] [PMC free article] [PubMed] [Google Scholar]
  62. Summerfield, C. , & de Lange, F. P. (2014). Expectation in perceptual decision making: Neural and computational mechanisms. Nature Reviews. Neuroscience, 15(11), 745–756. 10.1038/nrn3838 [DOI] [PubMed] [Google Scholar]
  63. Togoli, I. , Fedele, M. , Fornaciai, M. , & Bueti, D. (2021). Serial dependence in time and numerosity perception is dimension‐specific. Journal of Vision, 21(5), 6. 10.1167/jov.21.5.6 [DOI] [PMC free article] [PubMed] [Google Scholar]
  64. Wang, T. , Luo, Y. , Ivry, R. B. , Tsay, J. S. , Pöppel, E. , & Bao, Y. (2023). A unitary mechanism underlies adaptation to both local and global environmental statistics in time perception. PLoS Computational Biology, 19(5), e1011116. 10.1371/journal.pcbi.1011116 [DOI] [PMC free article] [PubMed] [Google Scholar]
  65. Wearden, J. H. (1997). Temporal bisection in humans with longer stimulus durations. Temporal Bisection in Humans with Longer Stimulus Durations, 4995(1), 24–44. [DOI] [PubMed] [Google Scholar]
  66. Wehrman, J. , Sanders, R. , & Wearden, J. (2023). What came before: Assimilation effects in the categorization of time intervals. Cognition, 234(105378), 105378. 10.1016/j.cognition.2023.105378 [DOI] [PubMed] [Google Scholar]
  67. Wehrman, J. J. , Wearden, J. , & Sowman, P. (2020). Decisional carryover effects in interval timing: Evidence of a generalized response bias. Attention, Perception & Psychophysics, 82(4), 2147–2164. 10.3758/s13414-019-01922-1 [DOI] [PubMed] [Google Scholar]
  68. Wei, X.‐X. , & Stocker, A. A. (2015). A Bayesian observer model constrained by efficient coding can explain “anti‐Bayesian” percepts. Nature Neuroscience, 18(10), 1509–1517. 10.1038/nn.4105 [DOI] [PubMed] [Google Scholar]
  69. Weiss, Y. , Simoncelli, E. P. , & Adelson, E. H. (2002). Motion illusions as optimal percepts. Nature Neuroscience, 5(6), 598–604. 10.1038/nn858 [DOI] [PubMed] [Google Scholar]
  70. Wiener, M. , Thompson, J. C. , & Coslett, H. B. (2014). Continuous carryover of temporal context dissociates response bias from perceptual influence for duration. PLoS One, 9(6), e100803. 10.1371/journal.pone.0100803 [DOI] [PMC free article] [PubMed] [Google Scholar]
  71. Zhang, Z. , & Lewis‐Peacock, J. A. (2024). Integration of history information drives serial dependence and stabilizes working memory representations. The Journal of Neuroscience: The Official Journal of the Society for Neuroscience, 44(32), e2399232024. 10.1523/JNEUROSCI.2399-23.2024 [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Figure S1.

BJOP-117-1396-s001.docx (583.6KB, docx)

Data Availability Statement

The data that support the findings of this study are openly available at https://github.com/Jiao2048/DurSeqEff_RprBscTasks.

Articles from British Journal of Psychology are provided here courtesy of Wiley

RESOURCES