Tracking millisecond time in the human brain: from sensory features extraction to perception

Visual events unfolding in milliseconds play an essential role in shaping our perceptual appraisal of the environment. In the human brain, the durations of such events are processed across different regions, from early visual to parietal and frontal areas. Previous functional Magnetic Resonance Imaging (fMRI) studies have shown that early visual areas respond to durations via monotonic tuning, while downstream areas employ unimodal Gaussian-like tuning. In my PhD work, I addressed unresolved questions about the implementation of these two tuning mechanisms (i.e., monotonic and unimodal) along the cortical hierarchy. For instance, to what extent are they linked to spatial processing in the visual stream? Are they hierarchically organized in the brain to eventually support a behavioral outcome? I addressed these questions in two 7-Tesla fMRI studies where I manipulated the duration and spatial position of visual stimuli and asked participants to perform a categorization task. Results show that duration tuning, from monotonic and spatially-selective, gradually becomes unimodal and spatially-invariant beyond early visual cortices. However, while in parietal and premotor regions unimodal tuning encodes for the duration at hand, in frontal cortices it encodes for the subjective boundary used to solve the task. Importantly, in the intraparietal sulcus, monotonic and unimodal duration tuning coexist, and temporal information gets integrated with spatial information by common selective responses. Overall, this work shows that changes in tuning properties of different brain areas along the cortical hierarchy mediate an abstraction process of temporal information: from sensory evidence, to quantity, to a percept.