Bayesian adaptive estimation of the sensory memory decay function: the quick Partial Report method
23.546, Saturday, 17-May, 8:30 am - 12:30 pm, Pavilion
Jongsoo Baek1, Luis Lesmes2, Zhong-Lin Lu1; 1Laboratory of Brain Processes (LOBES), Departments of Psychology, The Ohio State University, Columbus, OH, 2Adaptive Sensory Technology, Boston, MA
Sensory memory is the literal, modality-specific neural representation of sensory stimuli in the human brain. It provides the initial copy of external stimulation to human sense organs that can be processed by subsequent stages of perception and cognition. Recent studies suggest that sensory memory decays much faster for observers with mild cognitive impairment and may serve as an early sign of the Alzheimer’s disease. In the visual modality, iconic memory is best assessed with the partial report procedure. In this procedure, an array of letters appears briefly on the screen. A post-stimulus cue directs the observer to report the identity of the cued letter. Typically 600-800 trials are tested to measure the sensory memory decay function. The long testing time has prevented wide use of the test in clinical settings and special populations. Here we develop a quick partial report or qPR procedure based on a Bayesian adaptive framework to estimate the sensory memory decay function with much reduced testing time. Starting with a prior distribution of the parameters, the method selects the stimulus to maximize the expected information gain in the next trial. It then updates the probability distribution of the parameters based on the observer’s response by Bayesian inference. The procedure is iterated until the total number of trials reaches a set value. Simulation studies suggest that only 100 trials are necessary to reach accuracy of .47dB and precision of 2dB. The method was validated in a psychophysical experiment. Estimates of the sensory memory decay function obtained with 100 qPR trials showed good precision (SD = .55dB) and excellent agreement with those obtained with 1600 trials using the conventional procedure (mean RMSE = .48dB). qPR relieves the data collection burden in characterizing sensory memory and makes it possible to assess sensory memory in clinical settings and special populations.