%0 Journal Article %J eLife %D 2018 %T Neural dynamics at successive stages of the ventral visual stream are consistent with hierarchical error signals. %A Issa, Elias B %A Cadieu, Charles F %A DiCarlo, James J %K Animals %K Brain Mapping %K Face %K Humans %K Macaca mulatta %K Models %K Neurological %K Neurons %K Pattern Recognition %K Photic Stimulation %K Reaction Time %K Visual %K Visual Cortex %K Visual Perception %X

Ventral visual stream neural responses are dynamic, even for static image presentations. However, dynamical neural models of visual cortex are lacking as most progress has been made modeling static, time-averaged responses. Here, we studied population neural dynamics during face detection across three cortical processing stages. Remarkably,~30 milliseconds after the initially evoked response, we found that neurons in intermediate level areas decreased their responses to typical configurations of their preferred face parts relative to their response for atypical configurations even while neurons in higher areas achieved and maintained a preference for typical configurations. These hierarchical neural dynamics were inconsistent with standard feedforward circuits. Rather, recurrent models computing prediction errors between stages captured the observed temporal signatures. This model of neural dynamics, which simply augments the standard feedforward model of online vision, suggests that neural responses to static images may encode top-down prediction errors in addition to bottom-up feature estimates.

%B eLife %V 7 %8 11/2018 %G eng %U https://elifesciences.org/articles/42870https://cdn.elifesciences.org/articles/42870/elife-42870-v2.pdf %R 10.7554/eLife.42870 %0 Journal Article %J Behavioural Brain Research %D 2002 %T Receptive field structure in cortical area 3b of the alert monkey %A DiCarlo, James J %A Johnson, Kenneth O %K Action Potentials %K Afferent %K Animals %K Brain Mapping %K Evoked Potentials %K Haplorhini %K Models %K Neurological %K Neurons %K Orientation %K Reproducibility of Results %K Skin %K Somatosensory %K Somatosensory Cortex %X

More than 350 neurons with fingerpad receptive fields (RFs) were studied in cortical area 3b of three alert monkeys. Random dot patterns, which contain all stimulus patterns with equal probability, were scanned across these RFs at three velocities and eight directions to reveal the RFs’ spatial and temporal structure. Area 3b RFs are characterized by three components: (1) a single, central excitatory region of short duration, (2) one or more inhibitory regions, also of short duration, that are adjacent to and nearly synchronous with the excitation, and (3) a region of inhibition that overlaps the excitation partially or totally and is temporally delayed with respect to the first two components. As a result of these properties, RF spatial structure depends on scanning direction but is virtually unaffected by changes in scanning velocity. This RF characterization, which is derived solely from responses to scanned random-dot patterns, predicts a neuron's responses to random patterns accurately, as expected, but it also predicts orientation sensitivity and preferred orientation measured with a scanned bar. Both orientation sensitivity and the ratio of coincident inhibition (number 2 above) to excitation are stronger in the supra- and infragranular layers than in layer IV.

%B Behavioural Brain Research %V 135 %P 167 - 178 %8 01/2002 %G eng %U https://linkinghub.elsevier.com/retrieve/pii/S0166432802001626 %N 1-2 %! Behavioural Brain Research %R 10.1016/S0166-4328(02)00162-6