Dynamic Population Coding of Category Information in ITC and PFC

Ethan M. Meyers1,2, David J. Freedman3,4, Gabriel Kreiman2,5, Earl K. Miller1,3, Tomaso Poggio1,2

1. Department of Brain and Cognitive Sciences, MIT;   2. The McGovern Institute for Brain Research, MIT;   3.  The Picower Institute for Learning and Memory, RIKEN-MIT Neuroscience Research Center;  4.  Department of Neurobiology, The University of Chicago;  5.  Ophthalmology and Program in Neuroscience, Children's Hospital Boston, Harvard Medical School

Supplementary Figures

Figure S1:     All 42 stimuli that were shown during the experiment

Figure S2:     Accuracy levels for three different classifiers

Figure S3:     Illustration of how visual information can lead to high categorization decoding

Figure S4:     Supplementary data for the decoding of abstract category information

Figure S5:     Readout of ‘identity information’ using the best 2, 4, 8, 16, 32, 64, or 128 neurons

Figure S6:     Readout results of abstract category information excluding the “best” 1, 2, 4, 8, 16, 32, 64, and 128 neurons

Figure S7:     Identity information is also coding by changing patterns of neural activity

Figure S8:     Finer time course of abstract category information