Examining Reject Relations in Stimulus Equivalence Simulations

2025-06-30

AI Summary

This study investigates how reject relations interfere with the computational modeling of stimulus equivalence (SE) formation in artificial neural networks. Method: Using feedforward networks (FFNs), BERT, GPT, and a probabilistic associative baseline model, we conducted simulations under 18 training conditions—including linear sequences, one-to-many/multiple-to-one structures, and selection/reject relations—within a match-to-sample paradigm. Contribution/Results: Although incorporating reject relations and biased negative comparisons improved classification accuracy, no model significantly outperformed the probabilistic associative baseline; critically, Transformer-based models showed no evidence of genuine equivalence-class generalization beyond associative learning. This work provides the first systematic demonstration of the confounding effect of reject relations in SE modeling, arguing for stricter criteria to distinguish true equivalence-class formation from spurious associative learning. It offers methodological cautions and theoretical refinements for computational validation of SE.

Technology Category

• Reasoning under Uncertainty: Relational Probabilistic Models
• Cognitive Modeling & Cognitive Systems: Neural Spike Coding
• Knowledge Representation and Reasoning: Computational Complexity of Reasoning

Application Category

• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Economics, Online Markets and Human Computation: Social networks and social learning

Abstract

Simulations offer a valuable tool for exploring stimulus equivalence (SE), yet the potential of reject relations to disrupt the assessment of equivalence class formation is contentious. This study investigates the role of reject relations in the acquisition of stimulus equivalence using computational models. We examined feedforward neural networks (FFNs), bidirectional encoder representations from transformers (BERT), and generative pre-trained transformers (GPT) across 18 conditions in matching-to-sample (MTS) simulations. Conditions varied in training structure (linear series, one-to-many, and many-to-one), relation type (select-only, reject-only, and select-reject), and negative comparison selection (standard and biased). A probabilistic agent served as a benchmark, embodying purely associative learning. The primary goal was to determine whether artificial neural networks could demonstrate equivalence class formation or whether their performance reflected associative learning. Results showed that reject relations influenced agent performance. While some agents achieved high accuracy on equivalence tests, particularly with reject relations and biased negative comparisons, this performance was comparable to the probabilistic agent. These findings suggest that artificial neural networks, including transformer models, may rely on associative strategies rather than SE. This underscores the need for careful consideration of reject relations and more stringent criteria in computational models of equivalence.

Problem

• Role of reject relations in stimulus equivalence acquisition
• Impact of reject relations on computational model performance
• Differentiating associative learning from true equivalence in models

Innovation

• Used feedforward neural networks for simulations
• Applied BERT and GPT in MTS conditions
• Tested reject relations' impact on learning