Double responding: A new constraint for models of speeded decision making

Area of Science:

• Cognitive Psychology
• Computational Neuroscience
• Decision Science

Background:

• Evidence accumulation models (EAMs) are key for understanding speeded decision making.
• Existing EAMs often exhibit mimicry, hindering theoretical distinctions.
• A novel phenomenon, "double responding," is introduced to better constrain these models.

Purpose of the Study:

• To introduce and define "double responding" as a phenomenon to differentiate between EAMs.
• To test if accounting for double responding and choice-response time distributions can distinguish model predictions.
• To evaluate the performance of different EAM architectures, specifically those with lateral versus feed-forward inhibition.

Main Methods:

• Developed experimental paradigms to elicit and measure "double responding."
• Constrained various EAMs to simultaneously predict choice-response time distributions and double responding behavior.
• Compared the predictive accuracy of models incorporating lateral inhibition (e.g., leaky-competing accumulator) versus feed-forward inhibition (e.g., diffusion model).

Main Results:

• Double responding provides benchmarks that allow for the differentiation of EAM predictions.
• Models featuring lateral inhibition demonstrated accurate predictions for both choice-response time distributions and double responding.
• Models with feed-forward inhibition, such as the diffusion model, performed poorly under the extended framework.

Conclusions:

• Lateral inhibition provides a robust mechanism for accurate EAM predictions in speeded decision making.
• Feed-forward inhibition appears insufficient for capturing the complexities revealed by double responding.
• Further research on double responding is needed to definitively establish the best models of decision making.