Beinborn L., Hollenstein N. Cognitive Plausibility in Natural Language Processing

Springer, 2024. — xii, 158 p. — (Synthesis Lectures on Human Language Technologies) — ISBN: 978-3-031-43259-0, 978-3-031-43260-6.This book explores the cognitive plausibility of computational language models and why it’s an important factor in their development and evaluation. The authors present the idea that more can be learned about the cognitive plausibility of computational language models by linking signals of cognitive processing load in humans to interpretability methods that allow for the exploration of the hidden mechanisms of neural models. The book identifies limitations when applying the existing methodology for representational analysis to contextualized settings and critiques the current emphasis on form over more grounded approaches to modeling language. The authors discuss how novel techniques for transfer and curriculum learning could lead to cognitively more plausible generalization capabilities in models. The book also highlights the importance of instance-level evaluation and includes a thorough discussion of the ethical considerations that may arise throughout the various stages of cognitive plausibility research.Provides a resource for students and researchers in natural language processing, machine learning, and cognitive science.
Argues that model development targeted purely toward performance-driven evaluation has reached a ceiling.
Approaches the topic by focusing on behavioral patterns, representational structure, and procedural strategies.True PDF