Pith. sign in

REVIEW 4 major objections 4 minor 12 references

Recency effect disappears when information is integrated from independent perceptual sources

T0 review · 4 major / 4 minor · reviewed 2026-08-06 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read Recency effect disappears when evidence is integrated from independent perceptual sources.

desk verdict A clever new manipulation, but the claimed 'disappearance' of recency is not actually established by the reported statistics. read the letter →

arxiv 2506.21781 v1 pith:UMJQZZTA submitted 2025-06-26 q-bio.NC

Sepide Bagheri , Mehdi Keramati , Reza Ebrahimpour , Sajjad Zabbah This is my paper
classification q-bio.NC
keywords randomdotmotionperceptualdecision-makingdiscreteevidenceintegrationrecencyeffectsequencedependencysourceindependenceconfidenceadaptivegaincontrol
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper asks how people combine two brief, separated pulses of visual motion evidence, and whether the order of those pulses matters. Using a random-dot motion task, the authors show that the recency effect—better accuracy when the stronger pulse comes second—appears only when both pulses move in the same direction, and disappears when the two pulses move in orthogonal directions. Temporal gap between pulses does not affect either accuracy or confidence. The result matters because real decisions often mix information from distinct sources, and it suggests that the sequence bias found in earlier single-source experiments is not a universal feature of evidence accumulation.

What carries the argument

The key manipulation is source type: 'same' means the second motion pulse has the same direction as the first, while 'different' means the two pulses move in orthogonal directions within the same response class. The load-bearing analysis is logistic regression equation (6), which models correct choice as a function of pulse sequence, source type, and their interaction; the interaction term is what carries the claim that recency depends on source identity. The proposed mechanism is adaptive gain control: when the source of evidence stays constant, later evidence is processed with higher gain, producing a recency bias, whereas independent sources do not trigger that extra gain. This mechanism also explains why the effect disappears when evidence arrives from different spatial locations in earlier work.

What would settle it

Run the same two-pulse task while directly testing whether weak-strong and strong-weak accuracy differ within the different-source condition, or map the two orthogonal directions to separate response keys: if a recency effect reappears in either variant, the source-independence interpretation fails.

Watch

Extended reading notes

Core claim

The central claim is that the sequence-dependent accuracy effect in discrete evidence integration is specific to evidence coming from a single perceptual source. In same-source trials, accuracy is higher for weak-strong than strong-weak sequences, reproducing the known recency effect. In different-source trials, where the two random-dot pulses move in orthogonal directions, this ordering disappears. The authors support the claim with a logistic regression in which the interaction between pulse sequence and source type is significant (equation 6; $\beta_3 = 0.335$, $p = 0.016$), and they report that accuracy remains better than a perfect accumulator model for both source types. Confidence, like accuracy, is gap-independent, but is higher for different-source than same-source trials.

Load-bearing premise

The paper assumes that two orthogonal motion directions, both mapped to the same response key, count as independent perceptual sources; if that operational definition does not capture true source independence, the disappearance of the recency effect would not demonstrate what the authors conclude.

Editorial extensions

If this is right

  • If the claim is right, recency effects should shrink whenever evidence is gathered from genuinely separate perceptual sources, not just from separate pulses.
  • Models of evidence accumulation would need to track the identity of the source and not only the strength of each piece of evidence.
  • The observed independence of accuracy from temporal gap, across both source types, supports a stable accumulation process that does not decay over the tested intervals.
  • The higher confidence reported for different-source trials suggests confidence and choice are not driven by the same single accumulator.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • A natural extension is to vary the angular difference between the two pulse directions continuously; if source independence is the active ingredient, the recency effect should fade gradually as the directions move from identical toward orthogonal.
  • The same logic predicts that cross-modal or cross-location integration, where sources are clearly independent, should also lack the recency bias.
  • The confidence boost for different sources could be modeled as an additional source-uncertainty term in a metacognitive accumulator, a structure the current experiment does not distinguish.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

4 major / 4 minor

Summary. This manuscript reports a random-dot-motion study with four participants, examining how accuracy and confidence are affected when two discrete motion pulses come from the same direction versus orthogonal directions. The authors find that accuracy is independent of the temporal gap between pulses, that a recency effect (better accuracy when the second pulse is stronger) appears for same-direction pulses but is claimed to disappear for orthogonal-direction pulses, that accuracy exceeds a perfect accumulator model's predictions, and that confidence is higher for orthogonal sources but unaffected by sequence or gap. The paper includes a clearly described task, publicly available code and data, and an out-of-sample calibration of the perfect accumulator model.

Significance. If the disappearance of the recency effect for orthogonal sources were robust, it would extend the evidence-accumulation literature by showing that source consistency modulates sequence-dependent decision making. The study leverages a within-subject design, provides open code/data, and makes an explicit out-of-sample prediction. However, the headline claim is currently supported only by a pooled interaction term in a logistic regression, and the statistical evidence is not yet convincing enough to establish the claimed disappearance.

major comments (4)
  1. [§Results, 'Sequence-dependent behavior is influenced by source type'; Eq. (6), Table S7] The claim that the recency effect 'disappears' in the different-source condition is not established by the reported analysis. The significant interaction (β3 = 0.335, p = 0.016) only shows that the sequence effect differs between source conditions; it does not test whether the sequence effect is absent within the different-source condition. The manuscript reports no simple-effect slope of S for different-source trials, its confidence interval, or an equivalence test. Given that all four individual-level interaction estimates are non-significant (Sub1 p = 0.149, Sub2 p = 0.842, Sub3 p = 0.081, Sub4 p = 0.173), the claimed disappearance requires a direct within-condition null or equivalence test.
  2. [§Data analysis, Eq. (6), Table S7] Pooling all trials across the four participants for the logistic regression treats observations as independent even though they are clustered within subjects. With only four participants and no significant individual-level interaction, the pooled interaction (p = 0.016) may reflect between-subject heterogeneity rather than a consistent within-subject effect. A mixed-effects logistic model with random intercepts and slopes—or at least a subject-level summary analysis—should be used to support the cross-subject conclusion.
  3. [§Results, 'Regardless of source type, participants’ accuracy is higher than predicted by perfect accumulator model'…] The better-than-expected accuracy effect is not 'regardless of source type' according to the paper's own supplementary table. Table S10 shows Pe < Po only for Sub2 in the different-source condition and only for Sub2 and Sub4 in the same-source condition; Sub1 and Sub3 do not show the effect. The pooled β = 0.209 (p < 0.0001) masks this inconsistency, so the general claim is overstated and needs to be qualified or supported by a test that accounts for participant-level variation.
  4. [§Methods, 'Experimental task'] The binary response mapping treats U/R and D/L pairs, so 'different sources' are orthogonal directions that map to the same decision key. The manuscript does not independently validate that these directions constitute independent perceptual sources; they share a response category and may engage shared decision-level mechanisms. This weakens the conceptual interpretation of 'perceptual independency.' Please provide evidence that orthogonal directions within a response category are perceptually independent, or temper the wording.
minor comments (4)
  1. [Figure 1 legend] The word 'repose' should be 'response'; in the Supplementary Methods, 'mounted to' should be 'amounted to'.
  2. [Table S7, pooled row] The entry '-0.0.287' appears to be a typo for '-0.287'.
  3. [Table S10] The entries with '−' are unclear; please add annotation that these indicate no significant difference or that the test was not performed.
  4. [Abstract] The phrase 'perceptual independency' would be better rendered as 'perceptual independence'.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the central recency-disappearance claim is a direct empirical interaction test, and the perfect-accumulator comparison is an out-of-sample calibration, not a fitted-input prediction.

full rationale

This is an empirical behavioral study whose central claim (recency effect disappears when pulses come from orthogonal sources) is supported by a direct statistical interaction in Equation 6 (β3 = 0.335, p = 0.016; Table S7) and by the pattern in Figure 3b. That claim is not derived from a model that contains the conclusion by construction; it is a measurement of behavior. The perfect-accumulator prediction is built from single-pulse accuracy per participant and coherence level, then compared with observed double-pulse accuracy in Equation 8; this is a legitimate model-based benchmark rather than a fitted parameter renamed as a prediction, even though the same participants contributed to both. Self-citations (Tohidi-Moghaddam et al. 2019, Azizi & Ebrahimpour 2023, Majidpour et al. 2025) are used only to contextualize previously reported recency and gap-independence effects; the present manipulation and regressions carry the paper's own claims, so those citations are not load-bearing. The operational definition of 'independent perceptual sources' as orthogonal directions within one response category is a construct-validity concern, not a circular reduction, and the skeptical statistical objection (interaction alone does not prove absence of a simple effect) concerns inference strength, not circularity. No equation in the paper reduces to its own input by definition, and no fitted parameter is relabeled as a prediction. Therefore no significant circularity is present.

Assumptions & free parameters 0 free parameters · 4 assumptions · 0 invented entities

No mechanistic free parameters are introduced; the logistic regression coefficients are standard statistical estimates. The expected-accuracy model assumes Gaussian evidence with unit variance and additive integration, derived from single-pulse accuracy.

assumptions (4)
  • domain assumption Single-pulse accuracy can be converted to z-score evidence via the inverse cumulative normal
    Used to predict double-pulse accuracy: e = Φ^{-1}(P), then P_e = 1 - Φ(0 | μ=e1+e2, σ=√2).
  • domain assumption Evidence from the two pulses is independent and additive with Gaussian noise
    The perfect accumulator model formula assumes this to generate expected accuracy.
  • domain assumption The four motion directions can be collapsed into two response categories without loss
    Task design maps U/R to one key and D/L to another; all analyses pool across these categories.
  • domain assumption Adaptive gain control (Cheadle et al., 2014) explains recency and is sensitive to source consistency
    Invoked in the Discussion to interpret the source-type modulation; not directly tested here.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Recency effect disappears when information is integrated from independent perceptual sources." pith.science (2026). https://pith.science/paper/UMJQZZTA

@misc{pith2026250621781,
  author       = {Pith},
  title        = {Pith review of: Recency effect disappears when information is integrated from independent perceptual sources},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/UMJQZZTA}},
  note         = {Machine review of arXiv:2506.21781}
}
read the original abstract

Decision-making often involves integrating discrete pieces of information from distinct sources over time, yet the cognitive mechanisms underlying this integration remain unclear. In this study, we examined how individuals accumulate and integrate discrete sensory evidence. Participants performed a task involving random dot motion stimuli presented in single- and double-pulse trials. These stimuli varied in motion coherence, source consistency of pulses (either the same or orthogonal directions), and temporal gaps between pulses. We found that participants effectively integrated information regardless of source type or temporal gaps. As expected, when both pulses originated from the same source, performance showed a sequence-dependent effect-accuracy was influenced by the order of pulse presentation. However, this effect disappeared when the pulses came from orthogonal sources. Confidence judgments were similarly unaffected by temporal gaps or pulse sequence but were higher when information originated from orthogonal sources. These findings highlight the specific role of perceptual independency on information integration and consequently, on decision-making.

Figures

Figures reproduced from arXiv: 2506.21781 by the authors.

Figure 1
Figure 1. Direction discrimination task with single- and double-pulse random dot motions. In a single-pulse trial, the participant indicated readiness by pressing the spacebar on the keyboard after a red fixation point appeared at the center of the screen. Following this readiness response, a fixation gap of 200–500ms occurred, after which the red fixation point was replaced by a random dot stimulus displayed for 600ms (See S… view at source ↗
Figure 1
Figure 1. Direction discrimination task with single- and double-pulse random dot motions. In single-pulse trials, the motion was randomly chosen among four direction conditions (U, D, L, R) and six coherence levels (0%, 6.4%, 12.8%, 25.6%, 51.2%), and the delay was chosen from two conditions (500 or 1000 msec). In double-pulse trials, the first motion pulse was randomly chosen among four direction conditions (U, D, L, R), and… view at source ↗

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

12 extracted references · 10 canonical work pages

  1. [1]

    I., & Shadlen, M

    Gold, J. I., & Shadlen, M. N. (2007). The neural basis of decision making. Annual Review of Neuroscience, 30(1), 535–574. https://doi.org/10.1146/annurev.neuro.29.051605.113038

    work page arXiv 2007

  2. [2]

    Neural Comput 2008; 20 (4): 873–922

    Roger Ratcliff, Gail McKoon; The Diffusion Decision Model: Theory and Data for Two-Choice Decision Tasks. Neural Comput 2008; 20 (4): 873–922. doi: https://doi.org/10.1162/neco.2008.12-06-420

    doi:10.1162/neco.2008.12-06-420 2008

  3. [3]

    D., & Shadlen, M

    Kiani, R., Hanks, T. D., & Shadlen, M. N. (2013). Bounded integration in parietal cortex underlies decisions even when viewing duration is dictated by the environment. Journal of Neuroscience, 33(9), 3791–3800. https://doi.org/10.1523/JNEUROSCI.4241-12.2013

    work page doi:10.1523/jneurosci.4241-12.2013 2013

  4. [4]

    L., & Kiani, R

    Waskom, M. L., & Kiani, R. (2018). Decision-making through integration of sensory evidence at prolonged timescales. Current Biology, 28(23), 3850–3856.e9. https://doi.org/10.1016/j.cub.2018.10.021

    work page doi:10.1016/j.cub.2018.10.021 2018

  5. [5]

    Tohidi-Moghaddam, M., Zabbah, S., Olianezhad, F., & Ebrahimpour, R. (2019). Sequence-dependent sensitivity explains the accuracy of decisions when cues are separated with a gap. Attention Perception & Psychophysics, 81(8), 2745–2754. https://doi.org/10.3758/s13414-019-01810-8

    work page doi:10.3758/s13414-019-01810-8 2019

  6. [6]

    Kira, S., Yang, T., & Shadlen, M. N. (2015). A neural implementation of Wald’s sequential probability ratio test. Neuron, 85(4), 861–873. https://doi.org/10.1016/j.neuron.2015.01.007

    work page doi:10.1016/j.neuron.2015.01.007 2015

  7. [7]

    Azizi, Z., & Ebrahimpour, R. (2023). Explaining Integration of Evidence Separated by Temporal Gaps with Frontoparietal Circuit Models. Neuroscience, 509, 74–95. https://doi.org/10.1016/j.neuroscience.2022.10.019

    work page doi:10.1016/j.neuroscience.2022.10.019 2023

  8. [8]

    Majidpour, S., Sanayei, M., Ebrahimpour, R., & Zabbah, S. (2025). Better than expected performance effect depends on the spatial location of visual stimulus. Scientific Reports, 15(1). https://doi.org/10.1038/s41598- 024-82146-8

    doi:10.1038/s41598- 2025

Show all 12 references
  1. [9]

    Azizi, Z., Zabbah, S., Jahanitabesh, A., & Ebrahimpour, R. (2021). Improvement of association between confidence and accuracy after integration of separate evidence over time. bioRxiv (Cold Spring Harbor Laboratory). https://doi.org/10.1101/2021.06.20.449145

    2021 doi

  2. [10]

    Rahnev, D., & Denison, R. N. (2018). Suboptimality in perceptual decision making. Behavioral and Brain Sciences, 41. https://doi.org/10.1017/s0140525x18000936

    2018 doi

  3. [11]

    H., & Summerfield, C

    Cheadle, S., Wyart, V., Tsetsos, K., Myers, N., De Gardelle, V., Castañón, S. H., & Summerfield, C. (2014). Adaptive Gain Control during Human Perceptual Choice. Neuron, 81(6), 1429–1441. https://doi.org/10.1016/j.neuron.2014.01.020

    2014 doi

  4. [12]

    D” in repose to U or R directions, and “Z

    Wyart, V., Nobre, A. C., & Summerfield, C. (2012b). Dissociable prior influences of signal probability and relevance on visual contrast sensitivity. Proceedings of the National Academy of Sciences, 109(9), 3593–3598. https://doi.org/10.1073/pnas.1120118109 Competing interests ...

    2012 doi

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.