Trübutschek, D., Marti, S., Ueberschär, H., & Dehaene, S. (2019). Probing the limits of activity-silent non-conscious working memory. Proceedings of the National Academy of Sciences, 116(28), 14358–14367. https://doi.org/10.1073/pnas.1820730116
Interesting paper by a group from France about mental rotation in working memory for active vs. activity-silent working memories (i.e. AMIs vs. UMIs). This work is building on the empirical and theoretical work of UMIs, or “unconscious working memory” by which information can be retained without sustained active firing (and neural decodability) of that information. However, they pose the question of whether unconscious working memories can be manipulated in that state, or whether that informaiton needs first to be transformed from an unconscious to an active state prior to mental rotation. If so, then perhaps it’s a misnomer to refer to this memory state as “unconscious working memory”, but rather short-term memory – just storage, no manipulation.
The authors used a fast event-related spatial delayed recognion task. Participants briefly viewed (17 ms) a small target circle oriented somewhere along the radius of a larger circle. In the middle of a 3-s delay period, they were gien a retrocue to either rotate the target 120 deg clockwise, 120 deg counterclockwise, or not at all. Then, subjects reported the new location of the target, accoding to the manipulation instruction. Importantly, they also reported whether the target was seen or unseen on each trial. In a first experiment, they colleced behavior only. In a second experiment, they also recorded MEG while a new group performed the task.
Analyses focused on neural decoding differences between seen and unseen trials. For seen trials, they were able to deocode the “active” mainentance of the seen target image for most of the delay period. This signal was significantly different than on unseen trials.
The main result is shown in Fig. 5. First, on seen trials, the pre-rotation target location is decodable for nearly the entire delay-period, and the response location is decodable about half a second before the response. The transformation from the original to the new location is shown more cleary in Fig. 6, shown separately for clockwise, counterclockwise, and no rotations. These data are across trials, not within individual trials, and it shows (on average) a smooth transition between pre- and post-rotation locations. Second, on unseen trials, there is no sustained actiation of the actual unseen target location. However, based on the subects response on each trial, the authors computed a “pre-rotation location” which corresponds to where the target would be based on their respone. For example, if their response was 150 deg, and the cue was to rotation the (unseen) target clockwise (by 120 deg), then their pre-rotation location would be 150-120 = 30 deg. Now, attempting to decode this locaiton was successful during the pre-cue delay period.
The authors interpret this as indicating that subjects needed to transform an unconscious representation into a conscious/active one before applying the mental rotation to it. That is, the information could apparently not be manipulated in the unconscious state.
Overall, I found the motivation / logic of this study more compelling than their actual results. That is, the reference section is a strong list of recent work on “unconscious working memory” research. While the present results are a bit convulted and not totally convincing, especially in the ways the authors chose to present them. For example, in Fig. 6 showing the moment by moment decoding of location for seen trials, it would have been quite handy to see these same data for unseen trials. And it would have been handy to see some more analyses trying to link neural and behavior results – currently this is weak.
Overall, super relevant work to our lab. Great reference section that should be followed up. But the results here are a bit underwhelming. And their attempt to clarify the messy nomenclature in this are (e.g., “We therefore propose “activity-silent short-term memory” as an alternative term for the phenomenon of long- lasting blindsight.” is not helpful. So many terms floating around here. I prefer active vs. passive.
Signed, Jarrod Lewis-Peacock