Bromocriptine does not alter speed-accuracy tradeoff

Abstract

Being quick often comes at the expense of being accurate. This speed-accuracy tradeoff is a central feature of many types of decision making. It has been proposed that dopamine plays an important role in adjusting responses between fast and accurate behavior. In the current study we investigated the role of dopamine in perceptual decision making in humans, focusing on speed-accuracy tradeoff. Using a cued version of the random dot motion task, we instructed subjects to either make a fast or an accurate decision. We investigated decision making behavior in subjects who were given bromocriptine (a dopamine receptor agonist) or placebo. We analyzed the behavioral data using two accumulator models, the drift diffusion model, and the linear ballistic accumulator model. On a behavioral level, there were clear differences in decision threshold between speed and accuracy focus, but decision threshold did not differ between the drug and placebo sessions. Bayesian analyses support the null hypothesis that there is no effect of bromocriptine on decision threshold. On the neural level, we replicate previous findings that the striatum and pre-supplementary motor area are active when preparing for speed, compared with accurate decisions. We do not find an effect of bromocriptine on this activation. Therefore, we conclude that bromocriptine does not alter speed-accuracy tradeoff.

Keywords: bromocriptine; dopamine; drift diffusion model; functional magnetic resonance imaging; linear ballistic accumulator; model-based neuroimaging; speed–accuracy tradeoff; striatum.

Figure 1

The Drift Diffusion Model. Schematic illustration of the main components of the Drift Diffusion Model, showing a sample path for trial where a correct decision is made. Figure adapted from Mulder et al. (2012).

Figure 2

Summary statistics. Mean reaction times and accuracy rates across the two sessions (drug and placebo) and the two cues (speed and accuracy). Error bars indicate the standard error of the mean.

Figure 3

Model fits and parameter estimates. This figure shows the vincentized behavioral data and model fits (top) and the threshold estimates per experimental condition (bottom) of the LBA model (left) and the DDM (right). As the reaction time distributions in the top figure are plotted by their probabilities, the error distributions are shown on the left side of each plot, while the correct distributions are shown on the right side.

Figure 4

fMRI results. Results of the fMRI analyses showing the speed–accuracy contrast during placebo (yellow) and during drug (blue). (A) Cluster thresholded with a voxel thresholded of Z = 2.3, and a cluster threshold of p = 0.05. (B) Voxel thresholded results, with a voxel threshold of Z = 2.3 and no cluster threshold. The statistical maps are overlaid on Montreal Neurological Institute (MNI) T1 anatomical scans with a 2 mm resolution.