Decodability of short-term and long-term memory categories using fMRI

Abstract

Recent neuroscience focused on analyzing the spatial information of the brain as a code for its content. The multivariate analyses allow the brain signals to be decoded by classifying the categorical information in the distributed patterns of functional magnetic resonance imaging (fMRI) data. For a decade, many studies have been investigating the possibility of decoding the contents of the primary sensory information. Recent fMRI studies have suggested that even the memory, one of the highest cognitive functions, can be decoded as well though its underlying neural mechanism has remained unclear. How does the brain discriminate between perceived and memorized categorical information? In the present study fMRI based decoding technique to compare the decodability of re-imagined visual object in the absence of visual stimuli. To this end, visual short-term memory (visual STM) denotes the activity elicited in the brain immediately following the presentation of visual stimuli. In contrast‘ visual long-term memory (visual LTM) indicates the activity of the brain retrieved the information stored relatively longer before (i.e. 15 minutes following the offset of visual stimuli). This study investigated whether categorical information can be obtained from visual LTM and whether brain representations of visual STM and visual LTM are different using multi-voxel pattern analysis (MVPA).