GLAMbox

GLAMbox implements the Gaze-weighted Linear Accumulator Model (GLAM) to quantify how gaze allocation modulates evidence accumulation and choice behavior in multi-alternative decision tasks.

Key Features:

• Gaze-weighted Linear Accumulator Model: Assumes gaze-dependent evidence accumulation within a linear stochastic race framework that increases choice probability for items receiving more prolonged gaze.
• Bayesian Parameter Estimation: Supports Bayesian estimation of GLAM parameters at individual, pooled, and hierarchical levels.
• PyMC3-based Inference: Implements Bayesian statistical modeling using PyMC3 to fit GLAM to experimental choice and gaze data.
• Multi-alternative Support: Extends GLAM to choice sets with more than two alternatives.
• Choice Prediction: Generates predictions of choice behavior from gaze and value inputs.
• Individual-level Analysis: Characterizes variability in the gaze-choice association across participants.
• Data Visualization: Provides routines for visualization of choice- and gaze-related data outputs.

Scientific Applications:

• Gaze–choice association quantification: Quantifies how gaze allocation influences evidence accumulation and choice probabilities.
• Individual differences in decision processes: Characterizes participant-level variability in gaze-weighted evidence accumulation.
• Multi-alternative decision modeling: Applies GLAM to analyze decision-making in tasks with more than two alternatives.

Methodology:

GLAMbox implements the GLAM assumption of gaze-dependent evidence accumulation within a linear stochastic race framework and fits models using PyMC3 for Bayesian statistical inference, supporting individual, pooled, and hierarchical parameter estimation.