Geometry of sequence working memory in macaque prefrontal cortex

• Published in: Science (American Association for the Advancement of Science) Vol. 375; no. 6581; pp. 632 - 639
• Main Authors: Xie, Yang, Hu, Peiyao, Li, Junru, Chen, Jingwen, Song, Weibin, Wang, Xiao-Jing, Yang, Tianming, Dehaene, Stanislas, Tang, Shiming, Min, Bin, Wang, Liping
• Format: Journal Article
• Language: English
• Published: United States The American Association for the Advancement of Science 11.02.2022
• Subjects: Animals; Calcium - metabolism; Calcium imaging; Macaca mulatta; Mathematics; Memory, Short-Term; Models, Neurological; Neuroimaging; Neurons; Neurons - physiology; Prefrontal cortex; Prefrontal Cortex - cytology; Prefrontal Cortex - physiology; Saccadic eye movements; Short term memory; Spatial Memory; Subspaces
• ISSN: 0036-8075; 1095-9203; 1095-9203
• DOI: 10.1126/science.abm0204

How the brain stores a sequence in memory remains largely unknown. We investigated the neural code underlying sequence working memory using two-photon calcium imaging to record thousands of neurons in the prefrontal cortex of macaque monkeys memorizing and then reproducing a sequence of locations after a delay. We discovered a regular geometrical organization: The high-dimensional neural state space during the delay could be decomposed into a sum of low-dimensional subspaces, each storing the spatial location at a given ordinal rank, which could be generalized to novel sequences and explain monkey behavior. The rank subspaces were distributed across large overlapping neural groups, and the integration of ordinal and spatial information occurred at the collective level rather than within single neurons. Thus, a simple representational geometry underlies sequence working memory. How is serial order mentally encoded and stored in memory? Xie et al . addressed this question using two-photon calcium imaging to simultaneously record thousands of prefrontal neurons in monkeys performing a delayed visuospatial sequence reproduction task. The animals saw a sequence of three locations and, after a delay, made a saccade to the corresponding locations in the appropriate order. The data obtained using a mathematical decomposition technique support a new and simple type of model: a static population code with distinct and near-orthogonal subspaces for each rank in the sequence, all superimposed in the same overlapping groups of prefrontal neurons. These results open an important and new perspective on understanding sequence representations in the brain. —PRS Calcium imaging shows the representation of spatial working memory in primate lateral prefrontal cortex.