I did my secondment under the supervision of Pascal Mamassian at the Laboratoire des Systèmes Perceptifs (LSP) in Paris. I spent almost a month at the LSP investigating how speed uncertainty influences human motion perception under naturalistic conditions.
Specifically, we studied how the visual system integrates motion energy across different spatio-temporal channels to perform complex behaviors such as global motion perception under complex, noisy naturalistic conditions. It is also unknown how stimulus variability along a speed dimension evokes different perceptual responses. We addressed this question by manipulating local speed variability distribution (i.e. speed bandwidth) using a well-controlled class of broadband random-texture stimuli called Motion Clouds with continuous naturalistic spatiotemporal frequency spectra (Sanz-Leon et al., 2012; Simoncini et al., 2012).
Increasing variability along speed dimensions, observers experience different perceptual regimes. It is therefore challenging to probe perceptual states using classical, low-level tasks such as speed discrimination. We used a supra-threshold approach called Maximum Likelihood Difference Scaling (Knoblauch & Maloney, 2008, 2012; Maloney & Yang, 2003) experiment, with our speed bandwidth stimuli and other stimuli with multiple speed components, to investigate these different possible perceptual regimes.
MLDS is a technique used in psychology that allows the efficient estimation of underlying perceptual scales. It allows examining the whole range of stimuli and assessing how different levels of physical dimension map into different psychological levels of perception. In the experiment, observers are asked to judge perceptual similarity along the speed dimension of two pairs of moving MotionCloud with different speed bandwidth, a quadruple, and select the pair with larger differences.
We aim to identify different perceptual regimes within this space that correspond to motion coherency, motion transparency and motion incoherency. These results will allow us to further characterize the shape of the interaction kernels observed between different speed tuned channels and different spatiotemporal scales (Gekas et al., 2017) that underlies global velocity estimation.
Joining LSP, in addition to socializing with new people, inspired me novel research ideas related to my project. Receiving new ideas and suggestions helped me to make progress in my PhD work.