# How would a drift diffusion model explain effects in reaction time but not accuracy?

How would a drift-diffusion model explain a case where a variable has an effect in reaction time but not in accuracy. I know that one explanation would be a speed-accuracy tradeoff, in which the decision boundaries are so far apart that there are unlikely to be any incorrect answers. But, is there another way that doesn't involve such a trade-off?

This could happen if the variable in question affected only the nondecision time parameter -- denoted $$T_{er}$$ in the diagram below (from this paper) -- which adds a constant "waiting time" before the diffusion process begins, thus shifting the reaction time distribution to the right.

• Ah! Thank you! So you are saying that if the variable forced participants to pause before beginning the random drift, that could lead to such an effect? – Dave May 21 at 0:00
• Technically, yes. But keep in mind the psychological interpretation of the nondecision time parameter. It's intended to represent time spent on extremely low-level biological & perceptual processes taking place before it's even registered in a subject's brain that a stimulus has been presented. The nondecision times are on the order of tens of milliseconds. So for a stimulus that makes subjects "pause" because, say, it's complicated, it's not psychologically plausible (or at least not consistent with the spirit of the model) to say that this is due to effects on the nondecision time parameter. – Jake Westfall May 21 at 0:26