Fig. 6: Neural representation in the strategy-constrained model matches monkey data.

a Top: Anatomical location of the multielectrode arrays (Red—monkey B, Green—monkey Q, Blue—monkey S) superimposed on the 3D reconstructed brain of monkey S (IPS—intraparietal sulcus, STS—superior temporal sulcus, LF—lateral fissure). Example trials showing the performance of a linear decoder trained to estimate target distance from a population of simultaneously recorded neurons in left monkey posterior parietal cortex. Bottom: Like the top panel but decoded from the FPN activity of the strategy-constrained model (Model 4). b Fraction of variance in actual target distance explained by a linear decoder trained on the activity of monkeys’ PPC neurons (top) and the models’ FPN neurons (bottom). c Fraction of cumulative variance explained by the decoder as a function of the number of principal components used for decoding in the monkey data (top) and the models (bottom). Data for the average monkey is overlaid in the bottom panel for easy comparison. Error bars in (b) denote standard errors in the mean estimated by bootstrapping.