A model of dynamic, within-trial conflict resolution for decision making

Area of Science:

• Cognitive Psychology
• Neuroscience
• Computational Modeling

Background:

• Moment-to-moment fluctuations in visual attention raise questions about cognitive control dynamics.
• Traditional flanker tasks examine response inhibition, while models differ on attention's role (between-trial conflict vs. within-trial time-based).
• Existing sequential sampling models (SSMs) offer limited explanations for attention modulation during decision-making.

Purpose of the Study:

• To propose and test a novel hypothesis that within-trial attention and cognitive control are emergent properties of decision dynamics.
• To develop and compare SSMs with alternative assumptions about attention modulation and evidence accumulation.
• To investigate the interplay between attention, evidence accumulation, and decision-making processes.

Main Methods:

• Developed a set of sequential sampling models (SSMs) with varying attention (time- vs. control-based) and evidence accumulation (strong vs. weak correlation) mechanisms.
• Employed likelihood-free Bayesian approximation for quantified, subject-level model comparisons.
• Fit models to behavioral data from flanker task variants and electroencephalography (EEG) measures.

Main Results:

• Converging evidence across three experiments favored models incorporating control-based attention mechanisms.
• The best-fitting models utilized evidence accumulation mechanisms characterized by leak and lateral inhibition.
• These preferred models uniquely mapped onto observed decision-related brain signals (EEG).

Conclusions:

• Within-trial cognitive control and attention are emergent properties of the decision-making process itself.
• Control-based attention, coupled with leak-based evidence accumulation, provides a superior account of behavioral and neural data in flanker tasks.
• This framework advances our understanding of the dynamic interplay between attention, control, and decision computation.