Purchase this article with an account.
Joris Vangeneugden, Frank Pollick, Rufin Vogels; Functional differentiation of macaque visual temporal cortical neurons using a parameterized action space. Journal of Vision 2008;8(6):232. doi: https://doi.org/10.1167/8.6.232.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Neurons in the rostral superior temporal sulcus (STS) are responsive to visual displays of body movements (Puce and Perrett, 2003). We employed a parameterized action space to determine the representation of action similarity by visual temporal neurons and the contribution of form versus motion information to their responses. The action space consisted of whole body recordings of arm movements from an actor performing 3 different actions and 18 in-between blends of these 3 actions. We recorded from 240 responsive temporal neurons in fixating monkeys. Multidimensional scaling analysis showed that responses represented the similarity between the actions. Further tests indicated two classes of neurons: neurons responding as strongly to static presentations as to actions (“snapshot” neurons), and neurons not responding to static presentations, but responding to motion (“motion” neurons). Unlike many “snapshot” neurons, “motion” neurons still responded when presenting the arm or wrist-point in isolation. The “motion” and “snapshot” neurons were found predominantly in the upper bank/fundus and lower bank of the STS, respectively. Most “motion” neurons showed modulation of their response during the course of an action which correlated partially with the end-effector speed variations during an action. Further analysis of the responses of the “motion” neurons to temporally reversed action sequences showed that other kinematic factors beside speed must contribute to their action selectivity. The responses to the action sequences of most “snapshot” neurons correlated with their response selectivity for static snapshots. “Motion” neurons displayed an on average greater selectivity for these simple dynamic actions than “snapshot” neurons, and the former represented the parameteterized action configuration more faithfully than the latter. We speculate that the upper bank/fundus STS “motion” neurons code for visual kinematics and therefore contribute to coding of visual actions, while the lower bank STS (and IT) “snapshot” neurons can code for posture.
This PDF is available to Subscribers Only