Master 2 internship under the supervision of Laurent Perrinet

Orientation selectivity of neurons in the primary visual cortex (V1) is believed to be central to visual form perception. In many species (especially in mammals), early stages of visual form processing begin with orientation selectives neurons in V1.

Being simple and tracable, the orientation selectivity is one of the most investigated sensory feature in the mammalian brain. However, its underlying neural mechanisms are still not completely understood.

The main objective of this internship was to have a better understanding of the mechanisms implicated in orientation discrimination.

For this, we conducted a psychophysical study using synthetic oriented textures called Motion Clouds who model the distribution of orientation in natural visual scene. MC can be easily controlled by playing with different parameters and we decided to focus on 4 differents parameters:

• the orientation angle θ
• the orientation bandwidth Bθ
• the spatial frequency bandwidth Bsf
• the spatial frequency Sf

We shown participant MC and asked them to report the pattern orientation they perceived using the mouse.

Experiment apparatus:

Each participant have done 3 sessions of 750 trials. We observed that the orientation bandwidth, spatial frequency and spatial frequency bandwdith, all decreased the acuity to discriminate the orientation. This observations have already been reported in various publications but the negative impact of an increase in spatial frequency on orientation selectivity has not yet been observed.

This preliminary work could be served as a basis for a larger study on orientation discrimination. We could replicate this experiment with participants and/or play with differents parameters (luminance, constrast or duration)

It would be interesting to find out more about the behavior of primary visual cortex neurons when performing an orientation discrimination task by measuring their activity during this task using electrophysiological tools.

Another possibility would be to propose a new model of orientation selectivity that will based on deep-learning which performance will be evaluated with the same orientation discrimination task proposed before.

Observations and project's progress have been documented and presented in the form of a clear and concise technical report (Notebooks).

They have been posted weekly in the /Notebooks folder. The final codes are in /Notebooks/W18