Do Language Models Understand the Cognitive Tasks Given to Them? Investigations with the N-Back Paradigm
This work addresses the challenge of interpreting cognitive task results for language models, which is crucial for researchers in AI and cognitive science, though it is incremental in refining evaluation methodologies.
The paper investigates whether language models' poor performance on cognitive tasks like N-back is due to cognitive limitations or task comprehension issues, finding that task comprehension and maintenance are significant factors, with experiments showing performance declines as task difficulty increases up to 10-back.
Cognitive tasks originally developed for humans are now increasingly used to study language models. While applying these tasks is often straightforward, interpreting their results can be challenging. In particular, when a model underperforms, it is often unclear whether this results from a limitation in the cognitive ability being tested or a failure to understand the task itself. A recent study argues that GPT 3.5's declining performance on 2-back and 3-back tasks reflects a working memory capacity limit similar to humans (Gong et al., 2024). By analyzing a range of open-source language models of varying performance levels on these tasks, we show that the poor performance is due at least in part to a limitation in task comprehension and task set maintenance. We challenge the best-performing model with progressively harder versions of the task (up to 10-back) and experiment with alternative prompting strategies, before analyzing model attentions. Our larger aim is to contribute to the ongoing conversation around refining methodologies for the cognitive evaluation of language models.