Throughout our everyday routine we must make actions in the face of uncertainty. From a decision theoretic standpoint, optimal actions are those that maximize the value associated with the task. However, in order for humans to act optimally, it necessitates the brain has an accurate representation of both the reward and probability associated with each outcome. Previous research investigating how humans use value structure to perform reaching movements has exclusively focused on asymptotic performance, ignoring how this structure is learned. Therefore, my postdoctoral work with Ken Nakayama and Shin Shimojo explores how people learn and use target value to adjust their reach plans. We are currently writing-up this project, so please check-back soon for a description of these results.

In addition to understanding how value is learned to guide actions, my work with Colin Camerer, Antonio Rangel, Shin Shimojo, and Alice Lin investigates how expertise influences how the brain encodes value. We compared the brain activity of poker experts to poker novices when performing different decision theoretic tasks. We are currently writing-up this project, so please check-back soon for a description of these results.

Our perceptual system gathers information through several different sources (e.g., vision, audition, haptic - touch). Humans use these sources of information to recreate the scene from which the perceptual data was derived. This is done so effortlessly that it seems trivial; however, the ease with which we carry out this operation disguises its underlying complexity. For example, if we look at a cup of coffee on our desk and wish to grab it, we must turn the 2D visual input from our retina into a 3D representation of the cup and its surroundings. This is called the inverse vision problem, as there are an infinite number of 3D objects that can result from any given 2D image. Therefore, our brain is estimating the type and location of the objects in our environment. These estimates are not perfect - errors are introduced both by the insufficiency of perceptual information and neural processing. This makes the job of the brain to infer the state of nature in the presence of uncertainty, given the data and our prior knowledge.

Understanding how our brain uses knowledge about its uncertainty is a topic of interest in my research. Work I’ve done as a graduate student (with Paul Schrater) has demonstrated that the brain is aware of uncertainty in the visual information specifying a target’s location, and that this knowledge is incorporated into reach plans. More interestingly, we have also demonstrated that our brain is aware of a different type of uncertainty (i.e., coordinate transformation uncertainty - CTU) that stems from imperfect knowledge about the relative position of our body segments. These experimental and modeling efforts investigating CTU provide a possible interpretation for previously unexplained results in psychophysical and neuroscience literature.

Relevant Publications:

Schlicht, E.J., & Schrater, P.R. (2007). Impact of coordinate transformation uncertainty on human sensorimotor control. Journal of Neurophysiology, 97(6), pp. 4203-14.  

Schlicht, E.J., & Schrater, P.R. (In Press). Effects of visual uncertainty on grasping movements. Experimental Brain Research.  

In addition to uncertainty estimation, I am fascinated by how our brain selects and performs intelligent actions. Similar to the inverse vision problem, there are also an infinite number of ways to which we can reach and grasp an object. Despite this fact, people tend to reach and grasp objects in an extremely effortless and predictable manner. This suggests that the brain is using some sort of strategy (i.e., loss function) to reduce the ambiguity in the reaching task. Another endeavor of my graduate work has been to reverse-engineer a natural loss function that can predict the contact conditions (i.e., finger locations and velocities) people use when making grasping movements. The results from this effort suggest that people place their fingers in locations that minimize the force required for accurate object motion. More interestingly, this suggests that our brain is aware of the physics involved with object manipulation and uses that knowledge for reach planning.

Relevant Publications:

Schlicht, E.J. (2007). Statistical decision theory for human perception-action cycles. Ph.D. Thesis, University of Minnesota.  

Schlicht, E.J., Schrater, P.R., & Sloane, C.E. (In preparation). Influence of gravity on object-finger contact conditions during grasping movements. 

Schrater, P.R., & Schlicht, E.J. (2006). Internal models for object manipulation: Determining optimal contact locations, Technical Report TR 06-003, University of Minnesota.