Human Cognition
Human cognition research currently focuses on understanding the similarities and differences between human thought processes and those of large language models (LLMs), particularly exploring how LLMs handle uncertainty, stress, and complex reasoning tasks. Researchers utilize various deep learning architectures, including transformers, to analyze neuroimaging data (fMRI) and model human cognitive processes like attention, memory, and decision-making, often comparing LLM performance to human benchmarks across diverse cognitive tasks. This interdisciplinary field aims to improve AI systems by aligning them more closely with human cognition and, conversely, to gain a deeper understanding of the human mind through the lens of artificial intelligence.
Papers
Turing Representational Similarity Analysis (RSA): A Flexible Method for Measuring Alignment Between Human and Artificial Intelligence
Mattson Ogg, Ritwik Bose, Jamie Scharf, Christopher Ratto, Michael Wolmetz
Forma mentis networks predict creativity ratings of short texts via interpretable artificial intelligence in human and GPT-simulated raters
Edith Haim, Natalie Fischer, Salvatore Citraro, Giulio Rossetti, Massimo Stella
The Cognitive Capabilities of Generative AI: A Comparative Analysis with Human Benchmarks
Isaac R. Galatzer-Levy, David Munday, Jed McGiffin, Xin Liu, Danny Karmon, Ilia Labzovsky, Rivka Moroshko, Amir Zait, Daniel McDuff
Does Spatial Cognition Emerge in Frontier Models?
Santhosh Kumar Ramakrishnan, Erik Wijmans, Philipp Kraehenbuehl, Vladlen Koltun