Average activity, but not variability, is the dominant factor in the representation of object categories in the brain. (27th March 2017)

• Record Type: Journal Article
• Authors: Karimi-Rouzbahani, Hamid; Bagheri, Nasour; Ebrahimpour, Reza
• Is Part Of: Neuroscience. Volume 346(2017)
• Page Start: 14
• Page End: 28
• Publication Date: 2017-03-27
• Subjects: object category encoding -- signal variability -- EEG -- computational model -- representational analysis; Neurochemistry -- Periodicals; Neurophysiology -- Periodicals; Neurology -- Periodicals; Neurochimie -- Périodiques; Neurophysiologie -- Périodiques; Neurochemistry; Neurophysiology; Electronic journals; Periodicals; Electronic journals; 612.8
• DOI: 10.1016/j.neuroscience.2017.01.002
• Languages: English
• ISSNs: 0306-4522

Abstract

Highlights: Within-trial variability of neural activities provides information regarding object categories. The mean of activities plays a significant role in category encoding in the brain. The independent variability, although effective, contributes only moderately to the category representation. The correlated variability between the EEG channels has an ignorable effect on the representations. Abstract: To categorize the perceived objects, brain utilizes a broad set of its resources and encoding strategies. Yet, it remains elusive how the category information is encoded in the brain. While many classical studies have sought the category information in the across-trial-averaged activity of neurons/neural populations, several recent studies have observed category information also in the within-trial correlated variability of activities between neural populations (i.e. dependent variability). Moreover, other studies have observed that independent variability of activity, which is the variability of the measured neural activity without any influence from correlated variability with other neurons/populations, could also be modulated for improved categorization. However, it was unknown how important each of the three factors (i.e. average activity, dependent and independent variability of activities) was in category encoding. Therefore, we designed an EEG experiment in which human subjects viewed a set of object exemplars from four categories. Using a computational model, we evaluated the contribution of each factor separately in category encoding. Results showed that the average activity played a significant role while the independent variability, although effective, contributed moderately to the category encoding. The inter-channel dependent variability showed an ignorable effect on the encoding. We also investigated the role of those factors in the encoding of variations which showed similar effects. These results imply that the brain, rather than variability, seems to use the average activity to convey information on the category of the perceived objects.