Why do we make mistakes?

Our brains are imperfect computers. We systematically misperceive the world around us, misremember past events, and fail to make good choices even when we have all the necessary time and compute. A psychologist or behavioral economist might argue that mistakes are a byproduct of simplifying heuristics or that mistakes are caused by imperfections in psychological processes. By contrast, the noiseLab is working on biological explanations. We believe that understanding mistakes will be the key to understanding the fundamental limitations on neural processing, but also the evolution of intelligent behavior.

We are discovering that mistakes emerge from hardware constraints like bounds on neural firing rates (1), the dynamics of the autonomic nervous system (2), and species differences in brain architechtures (3). One consistent theme in our work is the idea that "mistakes" in one setting often create computational advantages in another (4, 5, 6)—a testament to evolution's capacity to generate surprising and elegant solutions when faced with competing objectives.

Our approach tightly integrates experiment and theory. Experimentally, we combine population-scale neuronal recordings, direct causal perturbations, and large-scale online psychophysics. Theoretically, we develop novel computational models of brain and behavior that naturally produce mistakes. We also leverage simulation to answer fundamental questions about the adaptive function(s) of mistakes. We are centered in neuroscience, but enthusiastically incorporate tools, approaches, and insight from other disciplines—especially ethology, the cognitive sciences, applied math, and artificial intelligence.