Captivated by thought: “Sticky” thinking leaves traces of perceptual decoupling in task-evoked pupil size

Throughout the day, we may sometimes catch ourselves in patterns of thought that we experience as rigid and difficult to disengage from. Such “sticky” thinking can be highly disruptive to ongoing tasks, and when it turns into rumination constitutes a vulnerability for mental disorders such as depression and anxiety. The main goal of the present study was to explore the stickiness dimension of thought, by investigating how stickiness is reflected in task performance and pupil size. To measure spontaneous thought processes, we asked participants to perform a sustained attention to response task (SART), in which we embedded the participant’s concerns to potentially increase the probability of observing sticky thinking. The results indicated that sticky thinking was most frequently experienced when participants were disengaged from the task. Such episodes of sticky thought could be discriminated from neutral and non-sticky thought by an increase in errors on infrequent no-go trials. Furthermore, we found that sticky thought was associated with smaller pupil responses during correct responding. These results demonstrate that participants can report on the stickiness of their thought, and that stickiness can be investigated using pupillometry. In addition, the results suggest that sticky thought may limit attention and exertion of cognitive control to the task.

While reading the manuscript, I feel that the authors imply that, finding smaller task-evoked pupil responses, automatically signals a perceptual decoupling process in action (e.g. lines 145,146, or 627-632). This seems like a form of reverse inference to me. The fact that perceptual decoupling (if it exists) might be linked to smaller pupils, does not necessarily mean that finding small pupils means that the brain is decoupling from the external environment. If this hypothesized process is really in act, we should observe other concurrent measures of reduced processing of the external environment: I don't think that the current paradigm in the study allows to do that, and additionally, there appears to be evidence for no differences in behavioral indexes of RT and RT variability between sticky and non-sticky thinking (e.g., line 623). While the stickiness factor did discriminate in no-go accuracy, this is a finding also common for mind-wandering thoughts in general, and is not intrinsically linked to perceptual decoupling. It is also not clear what the mechanism that would link smaller pupil task-evoked responses to a perceptual decoupling process would be. All in all, I do not think that the current study provides enough evidence to warrant strong affirmations such as those in lines 627-632, or lines 686-688. I don't know if the authors agree, but my suggestion would be to use a much more cautious language throughout the manuscript.
Some other comments that I have that I hope the authors will find useful: • Lines 58-84: This is true but obviously not limited to sticky thought, but is one of the main obstacles in the field of mind-wandering and of consciousness research in general. I feel like the whole paragraph could be shortened if the authors have an interest in doing so, given that thought probes are pretty much the standard way in the field to study these types of ongoing thoughts. • Lines 93-95: I have read the cited study and I couldn't find some of the results here described (e.g. off-task thought, but not stickiness specifically, seemed to be related to task accuracy). Could the authors double-check? My apologies in advance if I missed or misread something. • Line 97: One key reference that could be added here is Seli, Cheyne & Smilek (2013), JEP:HPP.
• Lines 141-142: This is slightly misleading as sticky thought is a novel concept. I do believe that there is some pupillometric research on rumination (e.g. Siegle, Steinhauser, Carter, Ramel & Thase, 2003), which could be added here, if the authors think it would be interesting for their argument. • Lines 147-149: I was confused by this sentence, as it is possible to dissociate, and measure separately, baseline to task-evoked pupil responses. Could the authors clarify? • Lines 149-151: Another study that could be discussed here and potentially in other parts of the manuscript is Konishi, Brown, Battaglini & Smallwood (2017), in which the authors find smaller baseline pupils for off-task thought, and specifically for negatively valenced and intrusive thoughts. As an author of that study I have a conflict of interest in pointing to it, but it seems to have some obvious links to the present manuscript (e.g. the concepts of intrusive and negative thoughts seems very close to those of rumination/worry). • Lines 163-165: This seems a little bit like a filler sentence. Maybe some concrete examples could be provided. • Line 166: a "which" seems to be missing in the phrase "in we embedded".
• Lines 170-174: To me it appears that these sentences construct a false dichotomy between sticky thoughts and general off-task thought/mind-wandering. For example, all the predictions described here can relate to off-task thought too. • Lines 182-183: How was the number of participants decided? • Lines 231-232: I think I missed how the experiment included 16 personal concern triplets, if the authors selected the 2 main concerns for each participant, and then translated each into a triplet of words. Sorry in advance if I misunderstood. • Lines 249-251: Was this second question always presented, even if for example, participants reported to be on-task or externally distracted in the first question? Were the authors still analysing such cases? • Figure 2: There's a small typo in the third question (maters instead of matters).
• Line 263: A reference could be added for PsychoPy (the most recent study is Peirce, Gray, Simpson, et al., 2019, Behavior Research Methods). • Line 304-305: It seems like the fixed order would cue participants to know in advance when a thought-probe would be presented. This could be seen as an issue, do the authors have any opinions on this? • Lines 322-324: How were these cut-offs decided?
• Lines 364-365: This decision seems a bit strange in the context of pupil size and arousal, as external distraction is likely arousing, the opposite of inalertness. • Lines 367-368: What was the reason? What was the point of including a 6 point scale if it's not used in the analyses? • Lines 481-483: Apologies in advance as I'm not sure if I've missed it, but was this result also taking into account the overall proportion of mind-wandering reports? If not, this result would be confounded by that, as they also increase over time. • Figures 5 and 6: The legend for the plot on the right could be a bit clearer (or is missing).
• Lines 511-512: There is a serious and common problem in these designs, in which a thought-probe is presented always after a target stimulus. Following a mistake, participants might confabulate and report they were not on-task. This is a hard problem to overcome, but I think that this possibility should be discussed, given that the whole field relies on self-reports. • Lines 516-519: Again, I am not 100% sure if I missed this, but was this analysis also taking into account the underlying proportion of mind-wandering thoughts? • Lines 534-535: Did the authors analyse baseline pupil sizes across attentional states with simpler models such as LMMs? Given the amount of previous research on this, I'm just wondering if the null result depends somewhat on the GAMMs. • Lines 565-567: It seems a bit strange that the difference is smaller between sticky and non-sticky than between sticky and neutral. If these are opposite of a continuous state, one would expect the difference to be bigger between sticky and non-sticky. Do the authors have any comments on this?
• Figure 10: The legend here could be clearer. Maybe it's better to use colored lines instead of different patterns. It could also be that it wasn't very clear because the image included in the review was lo-res. • Line 654: Minor typo, missing a "to" in "compared neutral/non-sticky".
• General: Would it be possible to have a table or a description of the best fitting models for each analysis conducted? • General: Did the authors measure and check the response times to the thought-probes? It sometimes happens that some participants will start to respond very quickly to those probes (because they want to finish the experiment faster). Such fast responses should be discarded, as it is debatable if the participants can introspect and report on their previous mental states so quickly.