Information Encoding in the Inferior Temporal Visual Cortex: Contributions of the Firing Rates and the Correlations Between the Firing of Neurons
NC Aggelopoulos,Edmund T. Rolls,Leonardo Franco
DOI: https://doi.org/10.1167/2.7.425
2010-01-01
Journal of Vision
Abstract:An important issue in visual neuroscience is the extent to which the neural code utilizes the numbers of spikes each neuron emits, or the relative time of firing of different neurons, which might reflect stimulus-dependent synchronization and thus could encode information. We analyzed the extent to which populations of primate inferior temporal visual cortex (IT) neurons utilize these different types of encoding by using a quantitative information-theoretic approach that compares the information encoded in these different ways (Panzeri et al, 1999). The formula quantifies the corrections to the instantaneous information rate which result from correlations in spike emission between pairs of neurons. The responses of small sets of neurons were simultaneously recorded with multiple single neuron microelectrodes while a rhesus macaque performed a visual fixation task and was shown 20 images effective for different IT neurons such as objects and faces. It was shown that almost all the information was present in the number of spikes emitted by each neuron, with stimulus-dependent synchronization effects adding for most sets of simultaneously recorded neurons almost no additional information. It was also found that the redundancy between the small sets of simultaneously recorded neurons was low, in the region of 4–10%. In addition, it was shown with a decoding procedure for measuring the information, that for the whole population of 20 neurons analysed, the information increased linearly with the number of neurons in the sample. The overall conclusion is that IT neurons convey information that is almost independent, with little redundancy; and that stimulus-dependent synchronization contributes very little to the code. The encoding is thus in an appropriate form for readout by receiving areas in which the neurons compute a dot product between the numbers of spikes received from different neurons and their synaptic weight vectors (see Rolls ET and Deco G, 2002, Computational Neuroscience of Vision, Oxford University Press; Panzeri S, Schultz SR, Treves A and Rolls,E.T., 1999, Correlations and the encoding of information in the nervous system. Proceedings of the Royal Society B 266: 1001–1012).