Sensorimotor learning compensates for irreducible perceptual bias
Curation statements for this article:-
Curated by eLife
eLife Assessment
This study examines how repeatedly tracking a tilted moving pattern with the eyes changes the smooth pursuit eye movements used to follow it. While the initial bias in pursuit direction evoked by such a stimulus does not change with training, later corrective smooth pursuit eye movements happen sooner. This is a valuable study that is of interest to researchers studying eye movements, motion perception, and sensorimotor learning. Yet, the strength of evidence is currently incomplete, as the possible influence of catch-up saccades is neglected and analytical choices are inconsistent; furthermore, alternative possible interpretations could be considered and discussed.
This article has been Reviewed by the following groups
Listed in
- Evaluated articles (eLife)
Abstract
Because of the uncertainties present in any sensory inflows, we experience perceptual biases that often contaminate sensorimotor transformations. How these biases can be prevented through perceptual learning or adaptation is an open question in inference theories of perception and motor control. We revisited this debate using a classic example of low-level perceptual bias, the aperture problem in motion perception and its consequences for smooth pursuit. In a visuomotor tracking paradigm, participants pursued left-, right-tilted, or upright Gabor patches moving horizontally. As expected, initial pursuit direction was biased toward the oblique directions. This initial pursuit error remained unchanged when participants repeatedly tracked a single tilted Gabor during training sessions conducted over several days. By contrast, post-training error correction became faster for the trained Gabor orientation, but not for its untrained, mirror-symmetric counterpart. This change in dynamics was explained by the emergence of a delayed compensatory pursuit response, best revealed when testing with an upright Gabor, a stimulus that would normally elicit unbiased tracking. This corrective eye movement was selective for both shape and motion features. It lagged by ∼40 ms after pursuit onset, suggesting that it was triggered by the biased motor command itself rather than by visual reafference. These results support a dissociation between low-level sensory computations, which appear resistant to modification over the timescale of our training protocol, and internal models of sensorimotor transformation, which can be rapidly updated to compensate for irreducible but predictable sensory-driven perceptual biases.
Article activity feed
-
eLife Assessment
This study examines how repeatedly tracking a tilted moving pattern with the eyes changes the smooth pursuit eye movements used to follow it. While the initial bias in pursuit direction evoked by such a stimulus does not change with training, later corrective smooth pursuit eye movements happen sooner. This is a valuable study that is of interest to researchers studying eye movements, motion perception, and sensorimotor learning. Yet, the strength of evidence is currently incomplete, as the possible influence of catch-up saccades is neglected and analytical choices are inconsistent; furthermore, alternative possible interpretations could be considered and discussed.
-
Reviewer #1 (Public review):
Summary:
Schoeffel and colleagues investigated the plasticity of sensorimotor biases. As a test case, they used the influence of the aperture problem on the initiation of smooth pursuit: when a grating is moving in an oblique direction relative to its orientation, pursuit initiation is biased towards the direction orthogonal to the orientation before this error is corrected and pursuit tracks the actual 2D motion direction. Here, this well-known phenomenon was studied before and after a training block that contained repeated exposure to one stimulus orientation. The initial pursuit response remained biased even after training, while an early correction of this bias increased after training. The correction only occurred for stripped patterns but not for a homogeneous target, indicating that the learning was …
Reviewer #1 (Public review):
Summary:
Schoeffel and colleagues investigated the plasticity of sensorimotor biases. As a test case, they used the influence of the aperture problem on the initiation of smooth pursuit: when a grating is moving in an oblique direction relative to its orientation, pursuit initiation is biased towards the direction orthogonal to the orientation before this error is corrected and pursuit tracks the actual 2D motion direction. Here, this well-known phenomenon was studied before and after a training block that contained repeated exposure to one stimulus orientation. The initial pursuit response remained biased even after training, while an early correction of this bias increased after training. The correction only occurred for stripped patterns but not for a homogeneous target, indicating that the learning was contingent on the stimulus. Since the correction occurred before sensory feedback of the initial pursuit response became available, the authors conclude that it is triggered predictively based on the motor signal.
Strengths:
The highlights of the study are the extensive data collection and the elaborate methods: the combination of pre- and post-tests within one session and training over five sessions allows for study plasticity at multiple time scales. The clever manipulation of stimulus features allows us to identify the reference frame of learning. The comprehensive and detailed analysis of smooth pursuit eye movements allows to reliably distinguish between the initial bias and its correction and to precisely assess the time course of correction, both within a trial and across trials.
Weaknesses:
The analysis is focused on smooth pursuit and does not consider catch-up saccades that were removed from the velocity traces by linear interpolation. Without information on the frequency and direction of catch-up saccades, it remains unclear whether saccades contributed to the correction and how they were affected by the training.
The authors interpret their results as evidence that the early pursuit correction is driven by the sensorimotor signal during initiation. I do not quite understand how they can ultimately exclude the possibility that the error correction is driven by a dynamic change in the sensory motion signal.
-
Reviewer #2 (Public review):
Summary of aims.
The authors ask whether training on an obliquely oriented Gabor patch changes how people later follow differently oriented Gabor patches with their eyes. Tilted patterns produce a bias in the initial direction of pursuit because of the aperture problem. The authors report that this initial bias does not change over five training sessions. They do report a compensatory pursuit response that corrects the direction of the eye movement, and that this correction becomes faster with training. They attribute the faster correction to a later compensatory response, and they interpret this as learning in internal sensorimotor models.
Strengths.
The experiment is well done, and the question of how sensorimotor learning deals with a perceptual bias that cannot be removed is interesting. The …
Reviewer #2 (Public review):
Summary of aims.
The authors ask whether training on an obliquely oriented Gabor patch changes how people later follow differently oriented Gabor patches with their eyes. Tilted patterns produce a bias in the initial direction of pursuit because of the aperture problem. The authors report that this initial bias does not change over five training sessions. They do report a compensatory pursuit response that corrects the direction of the eye movement, and that this correction becomes faster with training. They attribute the faster correction to a later compensatory response, and they interpret this as learning in internal sensorimotor models.
Strengths.
The experiment is well done, and the question of how sensorimotor learning deals with a perceptual bias that cannot be removed is interesting. The interpretations are reasonable and fit the general approach of the study. Including independent conditions and a separate cohort is also a useful addition.
Weaknesses.
My main concern is that the results are hard to follow and at times inconsistent, and the analyses may be more extensive than needed.
Do the results support the conclusions? The central idea, that training speeds up a delayed compensatory correction without changing the initial bias, is plausible and may well be correct. As the paper stands, however, the support is only partial. The inconsistent windows and sessions, the statistical issues, and the unclear description of how the key latencies were obtained make it hard to judge how robust the main findings are. The pattern of analyses can also give the impression that they were chosen after the fact.
Impact.
If the findings hold up under a clearer and more consistent analysis, they would be of interest to researchers studying eye movements, motion perception and sensorimotor learning.
-
Reviewer #3 (Public review):
Summary:
The study explored smooth pursuit dynamics when human participants tracked moving gratings and blobs. If the grating is oriented obliquely relative to the pursuit direction, the initial direction of smooth pursuit initiation reflects the well-known aperture problem and is orthogonal to the grating orientation (even if the grating itself is moving in another direction). Very quickly, pursuit then corrects itself because there is also the position signal indicating where the target is. The authors here trained participants to see only one orientation repeatedly. The initial aperture problem effect did not disappear in smooth pursuit initiation, but pursuit was faster to correct itself and follow the true target position. The authors interpret this as the pursuit system having a compensatory learned …
Reviewer #3 (Public review):
Summary:
The study explored smooth pursuit dynamics when human participants tracked moving gratings and blobs. If the grating is oriented obliquely relative to the pursuit direction, the initial direction of smooth pursuit initiation reflects the well-known aperture problem and is orthogonal to the grating orientation (even if the grating itself is moving in another direction). Very quickly, pursuit then corrects itself because there is also the position signal indicating where the target is. The authors here trained participants to see only one orientation repeatedly. The initial aperture problem effect did not disappear in smooth pursuit initiation, but pursuit was faster to correct itself and follow the true target position. The authors interpret this as the pursuit system having a compensatory learned mechanism.
Strengths:
The paradigm is suitable for exploring learning.
Weaknesses:
I have several issues with this study, which I summarize here and then further in the detailed comments below. First, an alternative explanation of the results could be sensory adaptation to the repeated exposure to the one grating orientation. That is, the repeated visual exposure to one grating might adapt the effect of the aperture problem, but the target position component of tracking remains strong because target motion was randomly interleaved. Second, the paradigm undoubtedly induced saccades. However, there was no analysis of them. The problem is that the saccades happen exactly at the same time at which the authors measure and infer "compensatory mechanisms" of smooth pursuit. And we know for a fact that saccades alter smooth pursuit parameters. So, are the authors measuring pursuit or measuring the effects of learning on saccades, and then the changes in the saccades themselves alter pursuit? This is a fundamental question, and it was a bit sad to see absolutely no mention of saccades at all in the whole paper! Third, for the effects with vertical gratings (in which no compensation should be observed but is observed), there could be alternative explanations for the observations, which are not considered at all. For example, the authors did not consider calibration asymmetries and also asymmetries in where people look (for extended objects like in the stimuli used, we know that there are biases in eye position that are very real). Finally, the paper is really not clear in many places, especially in the figures. It is leaving way too much for the reader to infer, and it needs to be written much more carefully.
-
-
-
-
Sensorimotor learning compensates for irreducible perceptual bias https://www.biorxiv.org/content/10.64898/2026.06.16.732722v1
-
Sensorimotor learning compensates for irreducible perceptual bias https://www.biorxiv.org/content/10.64898/2026.06.16.732722v1
-