Exogenous capture accounts for fundamental differences between pro- and antisaccade performance
Curation statements for this article:-
Curated by eLife
Evaluation Summary:
When the subjects are instructed to produce saccades away from suddenly appearing visual targets under time pressure, early saccades tend to be directed incorrectly to the peripheral target, suggesting that exogenous and endogenous signals that are related to the target position and instruction, respectively, compete to control the motor responses. In this study, the authors provide further evidence for the independence of these two processes by showing that they can account for temporal evolution of correct saccades regardless of the instruction, stimulus luminance or motor bias.
(This preprint has been reviewed by eLife. We include the public reviews from the reviewers here; the authors also receive private feedback with suggested changes to the manuscript. Reviewer #1 agreed to share their name with the authors.)
This article has been Reviewed by the following groups
Listed in
- Evaluated articles (eLife)
Abstract
To generate the next eye movement, oculomotor circuits take into consideration the physical salience of objects in view and current behavioral goals, exogenous and endogenous influences, respectively. However, the interactions between exogenous and endogenous mechanisms and their dynamic contributions to target selection have been difficult to resolve because they evolve extremely rapidly. In a recent study (Salinas et al., 2019), we achieved the necessary temporal precision using an urgent variant of the antisaccade task wherein motor plans are initiated early and choice accuracy depends sharply on when exactly the visual cue information becomes available. Empirical and modeling results indicated that the exogenous signal arrives ∼80 ms after cue onset and rapidly accelerates the (incorrect) plan toward the cue, whereas the informed endogenous signal arrives ∼25 ms later to favor the (correct) plan away from the cue. Here, we scrutinize a key mechanistic hypothesis about this dynamic, that the exogenous and endogenous signals act at different times and independently of each other. We test quantitative model predictions by comparing the performance of human participants instructed to look toward a visual cue or away from it under high urgency. We find that, indeed, the exogenous response is largely impervious to task instructions; it simply flips its sign relative to the correct choice, and this largely explains the drastic differences in psychometric performance between the two tasks. Thus, saccadic choices are strongly dictated by the alignment between salience and behavioral goals.
Article activity feed
-
-
Author Response:
We largely agree with the assessment of the Reviewers. Indeed, as noted by Reviewer #2, under the urgent conditions of our experiment, the onset of the cue modulates competing saccade plans that are already ongoing. The reviewer is correct in considering that the initial motor plans are endogenously generated, as they favor one location or the other based simply on the subject's internal bias or preference. We would just note that the endogenous signal that we focus on refers to a later modulation which, based on the perceived cue location and the task rules, directs the motor plans to the correct target location. According to our findings, this endogenous modulation occurs after the exogenous response and acts in the opposite way, boosting the anti-saccade plan and curtailing the activity that would otherwise trigger …
Author Response:
We largely agree with the assessment of the Reviewers. Indeed, as noted by Reviewer #2, under the urgent conditions of our experiment, the onset of the cue modulates competing saccade plans that are already ongoing. The reviewer is correct in considering that the initial motor plans are endogenously generated, as they favor one location or the other based simply on the subject's internal bias or preference. We would just note that the endogenous signal that we focus on refers to a later modulation which, based on the perceived cue location and the task rules, directs the motor plans to the correct target location. According to our findings, this endogenous modulation occurs after the exogenous response and acts in the opposite way, boosting the anti-saccade plan and curtailing the activity that would otherwise trigger an erroneous pro-saccade. Thus, three things may happen in each trial: (1) initial, uninformed motor plans are endogenously generated, (2) the cue onset exogenously reinforces the plan toward the cue, and (3) an informed endogenous signal suppresses the plan toward the cue and boosts the plan toward the anti location. We think the novelty here is in being able to characterize these distinct events, which unfold within a few tens of milliseconds of each other.
Reviewer #3 considered our conclusion that the exogenous response "is entirely insensitive to behavioral context" too strong, and that is a fair point. Conclusions apply to the degree that experimental conditions are valid in general, and furthermore, the deviations from the idealized predictions were small but not zero. However, we do not consider the assumption noted by the reviewer, that saccade-related neural activity ramps up before the saccade goal is known, as a weakness. We have, in fact, recorded such activity in several oculomotor areas using similar urgent-choice designs (Stanford et al., Nat Neurosci 13:379, 2010; Costello et al., J Neurosci 33:16394, 2013; Costello et al., J Neurophysiol 115:581, 2016; Scerra et al., Curr Biol 29:294, 2019; Seideman et al., bioRxiv, 2021, https://doi.org/10.1101/2021.02.16.431470), and the responses in the frontal eye field (FEF) in particular conform quite closely with those assumed by the model (Stanford et al., Nat Neurosci 13:379, 2010; Costello et al., 2013; Salinas et al., Front Comput Neurosci 4:153, 2010). Rather than a potential liability, we think the early ramping activity is a key constraint for any model of urgent choice performance.
-
Evaluation Summary:
When the subjects are instructed to produce saccades away from suddenly appearing visual targets under time pressure, early saccades tend to be directed incorrectly to the peripheral target, suggesting that exogenous and endogenous signals that are related to the target position and instruction, respectively, compete to control the motor responses. In this study, the authors provide further evidence for the independence of these two processes by showing that they can account for temporal evolution of correct saccades regardless of the instruction, stimulus luminance or motor bias.
(This preprint has been reviewed by eLife. We include the public reviews from the reviewers here; the authors also receive private feedback with suggested changes to the manuscript. Reviewer #1 agreed to share their name with the authors.)
-
Reviewer #1 (Public Review):
Previous work by this research group has demonstrated that during a brief period after the onset of a visual target in an ant-saccade task, saccades tend to be directed obligatorily towards the target before they can be redirected to the correct location. In this study, Goldstein et al. tested whether these separate exogenous and endogenous processes are independent by examining the temporal evolution of saccade accuracy in pro- and anti-saccade tasks in the same subjects. They found that the results were consistent with a pair of models that differ only in terms of whether the exogenous and endogenous processes eventually lead to the saccade in the same direction or not. In addition, the temporal change in accuracy was also parsimoniously explained by the same model even when there is a systematic motor …
Reviewer #1 (Public Review):
Previous work by this research group has demonstrated that during a brief period after the onset of a visual target in an ant-saccade task, saccades tend to be directed obligatorily towards the target before they can be redirected to the correct location. In this study, Goldstein et al. tested whether these separate exogenous and endogenous processes are independent by examining the temporal evolution of saccade accuracy in pro- and anti-saccade tasks in the same subjects. They found that the results were consistent with a pair of models that differ only in terms of whether the exogenous and endogenous processes eventually lead to the saccade in the same direction or not. In addition, the temporal change in accuracy was also parsimoniously explained by the same model even when there is a systematic motor bias resulting from the recent trial history. Overall, these results provide further evidence for the independence of exogenous and endogenous processes responsible for the control of saccades. The manuscript is written clearly, and there are no major concerns.
-
Reviewer #2 (Public Review):
For clinicians, an informal method of evaluating frontal lobe function is to hold up a finger and ask (tell) the patient to look in the opposite direction. It is claimed that healthy controls can easily look away, while individuals with frontal lobe damage/dysfunction cannot overcome an overwhelming desire to look at the finger. This paradigm, known as the "anti-saccade," offers a simple but useful method to measure, model, and investigate neural correlates of response inhibition. Consequently, there has arisen a vast literature on the subject, to which the current study contributes.
If this reviewer understands correctly, the main contribution of the current study is to show that the anti-saccade plan is not triggered by the visual cue. Rather, equal and opposite pro- and anti-saccade plans evolve in …
Reviewer #2 (Public Review):
For clinicians, an informal method of evaluating frontal lobe function is to hold up a finger and ask (tell) the patient to look in the opposite direction. It is claimed that healthy controls can easily look away, while individuals with frontal lobe damage/dysfunction cannot overcome an overwhelming desire to look at the finger. This paradigm, known as the "anti-saccade," offers a simple but useful method to measure, model, and investigate neural correlates of response inhibition. Consequently, there has arisen a vast literature on the subject, to which the current study contributes.
If this reviewer understands correctly, the main contribution of the current study is to show that the anti-saccade plan is not triggered by the visual cue. Rather, equal and opposite pro- and anti-saccade plans evolve in parallel but with some stochastic variability. That is, at any given time, one plan may be closer to threshold than the other. When the cue comes along, it reinforces whichever plan is spatially congruent with the cue, and inhibits the other plan. The movement that is eventually produced is simply determined by whichever plan reaches threshold first. Thus, anti-saccades only happen when the anti-saccade plan has, by chance, achieved an insurmountable advantage over the pro-saccade plan.
This seems like a plausible model and it clarifies that anti-saccades are not a change of plan, but a selection from multiple competing plans. It further clarifies that the movement plans (which are endogenously generated) are independent of the exogenous cue, and that nothing about the cue, other than its congruence with the underlying plans, signals whether the subject should produce a pro- or anti-saccade. The model is simple enough that it generates clear and testable neurophysiological predictions.
Given the size of the extant literature, the reviewer is at a disadvantage to evaluate the novelty or impact of this, work and therefore leaves this aspect of the evaluation to the experts. To a non-expert, there are a few places where the authors seem to be using novel terminology for conventional ideas. This practice should be discouraged. The main concern is that the report is written at a very technical level, as if meant only for true afficionados.
-
Reviewer #3 (Public Review):
Goldstein et al. use computational modeling and human psychophysics to try to untangle the roles of exogenous capture of attention, cognition, and motor planning. The authors have already made considerable progress on this issue, most notably in their eLife paper (Salinas et al. 2019). The authors adapted a race-to-threshold model to their urgent pro- and anti-saccade task (Fig. 2). This model, employing a straighforward 'flip' on the exogenouos signal during pro vs anti saccades, produces clear predictions regarding the behavioral tachometric curves (Fig. 3), including fine-grained predictions about the relative timing of shifts in performance (Fig. 3b). The authors go on to demonstrate that this model does, in fact, capture the behavioral dynamics (Fig. 4). The agreement between model predictions and …
Reviewer #3 (Public Review):
Goldstein et al. use computational modeling and human psychophysics to try to untangle the roles of exogenous capture of attention, cognition, and motor planning. The authors have already made considerable progress on this issue, most notably in their eLife paper (Salinas et al. 2019). The authors adapted a race-to-threshold model to their urgent pro- and anti-saccade task (Fig. 2). This model, employing a straighforward 'flip' on the exogenouos signal during pro vs anti saccades, produces clear predictions regarding the behavioral tachometric curves (Fig. 3), including fine-grained predictions about the relative timing of shifts in performance (Fig. 3b). The authors go on to demonstrate that this model does, in fact, capture the behavioral dynamics (Fig. 4). The agreement between model predictions and behavioral observations extends to individual subjects (Fig. 4c) and across stimulus luminance levels (Fig. 4e). Putative motor biases were likewise captured (across the population) by the model dynamics (Fig. 5), as well as the timing of individuals' performances across tasks (Fig. 6). The ability of the model to account for these various aspects of the data is impressive.
Strengths:
The authors have previously demonstrated a specialized task that seems to facilitate a principled analysis. The task itself is well-controlled and simple, like most experimental tasks, but it provides robust behavioral effects, lending support to the authors' claims.The manuscript is largely well-structured and clearly written throughout.
The behavioral results are clearly presented and the agreement between model and behavior is indeed remarkable.
Weaknesses:
First, it is not clear to me what the major advance is beyond previous work. This is not a commentary on the quality of the current work. Given the results, it is interesting to consider the authors' assumption that the exogenous response "is entirely insensitive to behavorial context". The results are compelling but seem to fall short of that broad (assumed) claim.There is an assumption that "saccade-related neural activity is ramping up" (Line 272) before the saccade goal is known. I understand the impetus for this reasoning, but I don't think it is the only (or maybe even, best) account of the phenomenon. It is unclear exactly how a motor plan, in its strictest neurobiological sense is prepared to an unknown location? In the current task, I imagine the argument would be that two rival motor plans are developed (i.e., Fig. 2) and the subject is effectively weighting their two guesses accordingly. But then that raises the question of whether the result is dependent on the constraints of the task. An alternative explanation is that the "readiness" to move is not purely motor and may be largely cognitive. This distinction is subtle and slippery, but it informs our understanding of the neural circuitry because, in this interpretation, ramping of saccade activity does not necessarily need to be true. I would have liked to have seen more discussion of the roles of motor preparation, visual-motor attention, and general arousal. These topics are often poorly defined and the authors' approach could help disambiguate the terms.
-
