This image shows what regions in a subject's brain were involved in a memory task.       Click  image  for more information.

   Undergraduate

Christopher Paynter

My Research at Carnegie Mellon University

My current project will apply my knowledge of biochemistry to psycho - pharmacology.  My advisor, Dr. Lynne Reder, has studied synthetic amnesia using a drug called midazolam, a benzodiazepine commonly used as an anesthetic in surgery that causes temporary anterograde amnesia as a side effect.  We believe the drug works by preventing new memory traces from being bound to context.  One earlier study done in our lab found impaired performance during a recognition memory task for subjects given midazolam injections compared with saline controls.  This impairment, however, only held for those classes of stimuli for which subjects could easily generate a distinctive label.[1] We intend to look at fMRI images of subjects’ brains during a recognition memory task to see what regions are specifically affected by the midazolam.  We are particularly interested in differential patterns of hippocampal activation, as this area has often been implicated in studies of contextual binding.  By linking the pharmacological, neurological, and behavioral data on this effect, we can achieve a much better understanding of memory.

I am also currently working on an ERP (event related potential) project involving problem-solving with Dr. Kenneth Kotovsky.*  When subjects are given a task called the balls-and-boxes puzzle,  they typically make a large number of correct moves in succession  near the end but are then unable to explain how they solved the puzzle, indicating some sort of non-conscious insight mechanism.  We  are giving subjects this task while measuring response-locked ERPs to  learn about the neural correlates of this non-conscious problem- solving mechanism.

My first project at Carnegie-Mellon built on my experience of using EEG methodology by finding the ERP  correlates of initial feeling-of-knowing.  This project sought in part to replicate a behavioral study conducted in our lab[2] which found that subjects were very good at quickly assessing whether they would be able to retrieve the answer to a problem.  This assessment was made in a brief time window (under 850ms) that was much shorter than the time required to actually retrieve the answer itself.  This rapid, metacognitive judgment was shown to use a heuristic based on the familiarity of terms in the question, not the availability of the answer per se.  We believe that this heuristic enables efficient and flexible strategy selection during cognitive tasks. The study we just completed built on this work by adding an EEG component.  In it, subjects were presented with a series of previously unfamiliar math problems, some of which repeated over the course of the experiment.  Upon first seeing the problem, subjects rapidly decided within 850 milliseconds whether to solve it by retrieving the answer from memory or calculating it on scrap paper.  Subjects had 25 seconds to type the answer if they choose “calculate,” but only 2 seconds if they choose “retrieve.”  A large bonus was given for successful retrieves so subjects had an incentive to learn the answers.  For the EEG analysis, we were particularly interested in differences in brain wave patterns in the initial strategy selection phase.  Patterns unique to retrieve trials during the first 850 milliseconds reflected those processes relevant to initial feeling-of-knowing.  More specifically, we  looked at differential patterns of theta wave activity, an important marker of hippocampal activation, as well as the P300 component, a brainwave pattern indicating recognition of a stimulus as task-relevant.  Previously unfamiliar math problems were repeatedly tested over the course of the experiment and the feeling of knowing or not-knowing judgment had to be completed in 850 milliseconds (much less time than needed to retrieve the answer itself). ERP analyses uncovered waveform differences between accurate retrieve vs. calculate trials as early as 200 ms following onset of the problem.  Accurate retrieve trials showed activation primarily in the right hemisphere following problem onset, while calculate trials and inaccurate retrieve trials showed activation primarily in the left hemisphere. I submitted the results of our study for publication to Neuropsychologia.  I also presented a poster of our results at the Psychonomic Society's Annual Meeting in Long Beach, CA in November 2007.