Elizabeth Gifford
egifford at brandeis dot edu

Nothing is really work unless you would rather be doing something else.   - James M. Barrie
Vision Lab
Honors Thesis
Vision Lab

Slides from IGERT project 2004

My research experience began as a research assistant in the Vision Lab at Brandeis University in the fall of 2003. I began my tenure there by assisting graduate researchers in project analysis, working through transcripts of testing sessions and cataloguing data for further analysis by lead researchers. By the spring semester, I was working under the guidance of the PI conducting a follow-up experiment to a project conducted by a former graduate of the lab. The PI wanted to investigate whether there was occurrence of hysteresis in repeated trials with varied auditory and visual stimuli. During that semester I worked with lab members to develop a sufficient test for hysteresis, updated old code from the original experiment to implement our new experiment design, ran subjects through trials and did a limited amount of data analysis. We did not find significant evidence of hysteresis in the effects of repeated exposure to stimuli, but I had the opportunity to learn about experiment design and human subject testing.

The following summer, I received a position in the IGERT Summer Research Program to continue research in the same lab. My project, “Configural Information in Episodic Visual Recognition,” aimed to discover whether an inter-stimulus relation within a series of abstract stimuli enabled subjects to recall them more easily. I worked cooperatively with other lab members to develop the experiment design, particularly with one collaborating research partner, but was the lead researcher on the project. Our test stimuli were compound gratings with varying frequencies. Before the experiment, each subject was given a threshold test to determine the least noticeable difference in frequencies for each subject. The experiment trials were then scaled to fit a particular subject’s threshold. This was necessary to ensure that each subject was experiencing as close to the same trials as possible, despite differing abilities in distinguishing spatial frequency. In each trial a subject was presented with a series of three gratings, and then asked whether a fourth grating had been present in that series. Target trials did include the test stimulus in the series, while lure trials did not. Trials were arranged into two blocks. A type blocks consisted of mathematically related stimuli (where the change in frequency followed a linear progression through the trial), while B-type blocks consisted of the same stimuli as the A block, but 80 percent of trials had been re-ordered so as not to be sequential. Monetary compensation was given for time spent in testing sessions, and a bonus was given for high percentages of correct answers as an incentive for accuracy. We did not find any significant difference between the overall performance between Block A and B trials. We did find that in block B trials that were not arranged linearly subjects were able to more accurately recognize the second stimulus presented in a trial, while subjects performed better in recognizing the first or third stimulus in the trials that did contain linear progression. However, we also noted that the overall accuracy was relatively low and that subjects had low sensitivity to signal in both the linear and non-linear trials. We attributed this to not having enough difference between the lures and targets in trials due to the limited frequency range available to us. At the end of the summer, I gave a 30-minute presentation on my findings. The most useful skills I developed were using unexpected results to learn about the phenomenon we were studying and diagnosing problems in an experiment based on the results.

In a related study, I designed, programmed and executed an experiment to determine whether a subject’s perception of similar space is isomorphic with physical space. On each trial, subjects were presented with three complex gratings arranged equidistance from one another. Subjects were asked to select the most similar and most different pairs of gratings from the set. The three pairs of stimuli were then given similarity rankings based on the subject's judgment: the highest ranking was given to the most similar pair, lowest rank to the least similar, and by default the median ranking was given to the pair not deemed to be either the most similar or most different. The subject-defined similarities were then compared to the physical similarities of the stimuli. Early results showed that subjects put a greater emphasis on the similarity between gratings of higher frequencies than the lower frequency stimuli, which does not correspond to the consistent physical similarity between all of the gratings in this experiment.



Made on a Mac