Lingo
← Back to Neuroscience

Leaky Competing Accumulator (LCA)

Overview

The Leaky Competing Accumulator (LCA) model, developed by Usher and McClelland (2001), offers a neurally plausible framework for human decision-making. Like race models, it uses separate accumulators for each choice alternative. However, it incorporates two critical biologically inspired mechanisms: leakage and lateral inhibition.

Core Mechanisms: Leak & Inhibition

Unlike standard accumulation models where evidence builds up indefinitely, the LCA assumes that neural representations decay over time.

  • Leakage (λ): Accumulated evidence decays proportional to its current activation. This prevents runaway activation and naturally models forgetting or loss of context.
  • Lateral Inhibition (β): Accumulators actively suppress each other. As one option gains evidence, it suppresses competing options, acting as a competitive "winner-take-all" mechanism.

Neural Plausibility

The LCA is highly influential because it maps directly onto neurophysiological findings. Recordings in the posterior parietal cortex and frontal eye fields during choice tasks show precisely these dynamics: recurrent excitation (accumulation), decay (leak), and mutual suppression between neural populations encoding different targets.

References

  1. Usher & McClelland (2001) — Psychological Review— The foundational paper detailing the LCA model and its neural inspiration.

Examples