Summary of article (Milosavljevic)

The Drift Diffusion Model can account for the accuracy and reaction time of value-based choices under high and low time pressure

(Milosavljevic, Malmaud, Huth, Koch, & Rangel, 2010)

The Drift Diffusion Model (DDM) has spawned several versions. This article discusses the different versions in the context of a food-based experiment in an attempt to determine which of the different diffusion models explain the phenomenon under question under high and low time pressure.

The DDM has proved its efficacy in perceptual decision making. This study applies it to value-based decision making. The authors attempted to determine whether the DDM could account for accuracy and reaction time (RT) in food choices. Two different time-pressure studies are undertaken to see if difference in accuracy and RT occurs.

  1. The seven-parameter version of the Ratcliff DDM model had the best results according to their findings. Other DDM models used were:
  2. The Simple DDM (sDDM)
  3. The full DDM (fDDM). This provide a more accurate description of reaction time distributions than the sDDM and consists of 8 parameters
  4. Simple collapsing barrier DDM (scbDDM) where the height of the upper and lower decision thresholds reduce exponentially with time
  5. Full collapsing barrier DDM (fDDM) which like the scbDDM suggests that the decision thresholds reduce with time, but additionally measures the rate of reduction

Algorithms are provided for all these models.

Method:

Eight subjects participated in two experiments of one hour each. The first experiment consisted of a liking rating task where subjects were presented on screen with a series of pictures of (single) food items while at the same time right-eye tracking was used. They were asked to rate their liking of the food on a discrete scale of -2 to 2. This provided an individual subjective rating of the food items that was used in the next experiment. In the second experiment, subjects were asked to choose their preference from pairs of food items after the food items were presented on the screen in pairs for 2ms. Choice was captured by fixation of eye movements on the area of the screen where their first choice was.

In order to test the applicability of the different models, there were two different task conditions – time pressure and no time pressure. The results obtained suggested that the seven-parameter version of the DDM can account for the data with high-accuracy in both the high and low time pressure conditions.

MY NOTES:

  1. Gives a good outline of the pros and cons of the drift diffusion model.
  2. Value-based choice involves stimuli associated with reward.

Milosavljevic, M., Malmaud, J., Huth, A., Koch, C., & Rangel, A. (2010). The Drift Diffusion Model Can Account for the Accuracy and Reaction Time of Value-Based Choices Under High and Low Time Pressure. SSRN Electronic Journal, 5(6), 437–449. http://doi.org/10.2139/ssrn.1901533