author image

anna [ dot ] leshinskaya [ at ] gmail.com

CV

Anna Leshinskaya

Cognitive Scientist

I am a cognitive scientist, neuroscientist, and AI researcher. I am interested in the computational principles underlying the learning and representation of concepts and world models. I examine this in human behavior, human neural representation, and in AI systems. How do we use individual experiences to build what we know about the world? What is the representational structure of this knowledge? What is the neural architecture supporting this? I have published widely in human cognitive science and neuroscience. More recently, I have begun to leverage insights from human cognitive science to better understand the nature of cognition in large language models. I lead a research program in AI alignment at the AI Objectives Institute, and you can read more about that work in this blog post.

Research Highlights

blog image

Tools as Causers

What Drives Tool- and Hand- Selective Cortex?

An area in lateral temporal cortex (LOTC) responds preferentially to images of real-world tools and hands relative to other objects. What drives this selectivity? By teaching participants about novel objects, we gained an important insight: we found that novel objects that appear to influence/cause other events engage this area more than objects which move in response to events. We argue that a causal schema -- beyond shape or motor association -- is an important driver of the selectivity in LOTC and other parts of the temporal lobe. Read the pre-print.

blog image

Relations & Features

The Emergence of Relational Representation

How do you represent that coffee goes with bagels, but also looks like tea? Using newly learned relations among novel stimuli, we find that these kinds of representations are localized in distinct parts of the lateral temporal lobe. What things look like, and what they are associated with, are represented distinctly. This was published in Cerebral Cortex.

blog image

Memory Reflects Causal Inference

"Backward Blocking" effects in statistical learning

When we incidentally observe pairs of events, say clouds and rain, we bind them in associative memory, thinking of one when we see the other. However, what we recall as related is not just a matter of whether things appeared together: we are also influenced by whether those conjunctions are unique. Just like in causal reasoning, you wouldn't think clouds and rain were related if you often saw rain without clouds. We found that these considerations also govern what people remember from an incidental learning task. Check out the pre-print.or the Cog Sci paper.

Publications

Leshinskaya, A. & Chakroff, A. (2023). Value as semantics: representations of human moral and hedonic value in large language models. Advances in Neural Information Processing Systems (37), AI meets moral philosophy and moral psychology workshop. PDF

Leshinskaya, A., & Thompson-Schill, S.L. (under review). Computations of contingency guide how experience is encoded in memory: backward blocking in statistical learning. psyArXiv

Leshinskaya, A., Nguyen, M.A., & Ranganath, C. (2023). Integration of event experiences to build relational memory in the human brain. Cerebral Cortex. PDF

Leshinskaya, A., Bajaj, M. & Thompson-Schill, S.L. (2023). Novel objects with causal schemas elicit selective responses in tool- and hand-selective lateraloccipito-temporal cortex. Cerebral Cortex, 33 (9), 5557-5573. pdf

Leshinskaya, A., & Lambert, E. (2022). Implications from the philosophy of concepts for the neuroscience of memory systems. In F. De Brigard & W. Sinnott-Armstrong (Eds.), Neuroscience and Philosophy. Cambridge, MA: MIT Press.

Leshinskaya, A., Bajaj, M. & Thompson-Schill, S.L. (2020). Incidental binding between predictive relations. Cognition. 199(2020):104238.pdf

Leshinskaya, A., & Thompson-Schill, S.L. (2020). Transformation of event representations along middle temporal gyrus. Cerebral Cortex. 30(5), 3148–3166. pdf

Leshinskaya, A., & Wurm, M. F., & Caramazza, A. (2020). Concepts of actions and their objects. In M. Gazzaniga, G. R. Mangun, & D. Poeppel. The Cognitive Neurosciences, 6th edition. pdf

Leshinskaya, A., Lambert, E., & Thompson-Schill, S.L. (2019). Algebraic patterns as ensemble representations. In A.K. Goel, C. M.Seifert, & C. Freska (Eds.), Proceedings of the 41st Annual Conference of the Cognitive Science Society (pp. 646-650). Montreal, QB: Cognitive Science Society. pdf

Leshinskaya, A., & Thompson-Schill, S.L. (2019). From the structure of experience to concepts of structure: how the concept 'cause' applies to objects and events. Journal of Experimental Psychology: General 148 (4), 619-643. pdf

Leshinskaya, A., & Thompson-Schill, S.L. (2018). Inferences about uniqueness in statistical learning. In T.T. Rogers, M. Rau, X. Zhu, & C. W. Kalish (Eds.), Proceedings of the 40th Annual Conference of the Cognitive Science Society (pp 2020-2025) Austin, TX: Cognitive Science Society. pdf

Leshinskaya, A., Contreras, J.M., Caramazza, A. & Mitchell, J.P. (2017). Neural representations of belief concepts: A representational similarity approach to social semantics. Cerebral Cortex, 27, 344–357. pdf

Leshinskaya, A., & Caramazza, A. (2016). For a cognitive neuroscience of concepts : Moving beyond the grounding issue. Psychonomic Bulletin & Review, 23(4), 991–1001. pdf

Leshinskaya, A., & Caramazza, A. (2015). Abstract categories of functions in anterior parietal lobe. Neuropsychologia, 76, 27–40. pdf

Leshinskaya, A., & Caramazza, A. (2014). Nonmotor aspects of action concepts. Journal of Cognitive Neuroscience, 26(12), 2863–2879. pdf

Leshinskaya, A., & Caramazza, A. (2014). Organization and structure of conceptual representations. In V. Ferreira, M. Goldrick, & M. Miozzo (Eds.),Oxford Handbook of Language Production (pp. 118–133). Oxford: Oxford University Press. pdf