Rachel Lerch, RPI Graudate Student

 

Rachel Lerch, RPI Graudate Student

Sage 4101

March 7, 2018 12:00 PM - 1:30 PM

Perception and perceptual working memory are central to daily life. Even basic tasks, such as comparing the color or firmness of two objects, require the storage and manipulation of perceptual information in working memory. However, given the limitations of biological information processing, it is necessary that some information be lost or thrown away in the act of perceptual processing. This perspective ties directly to a branch of information theory known as rate–distortion theory (RDT), which provides the mathematics of optimal, but “lossy” information compression. When viewed from this perspective, the goals of engineered signal communication align with those of biological systems, in that they both are concerned with the minimization of communication error given constraints on capacity. Here, I present work on two experiments that apply the mathematical framework of RDT to the study of working memory for visual and haptic perceptual information.  These two rather exploratory studies are used as a proof of concept towards more complex applications. In particular, future work to extend RDT to the domain of reinforcement learning —optimizing behavior from trial-and- error feedback — will also be briefly discussed.

 

 

 

 

 

Add to calendar
Share|