CCN 2026 | Community Event: Toward a unified science of perception and memory

Value of the Information & Strength of the Argument

The value of the information is high, as it synthesizes recent empirical findings and computational approaches from leading researchers in the field. The argumentation is solid, grounded in experimental data (e.g., electrophysiology, fMRI, lesion studies) and computational modeling. The speakers present a coherent narrative that challenges traditional views of perception and memory as separate processes, and they provide concrete evidence for the role of MTL in integrating sequential visual information. The discussion is open and acknowledges limitations, such as the divergence between neural representations and behavior, which strengthens the credibility of the arguments.

Scientific Rigor, Source Quality, Title Accuracy

The scientific rigor is high, with references to published studies and ongoing research. The sources cited include the conference website and the speakers’ own work, though specific citations are not detailed in the transcript. The title accurately reflects the content, which focuses on unifying perception and memory research. The event is well-structured and the speakers are experts in their fields, contributing to the overall reliability of the information.

The title accurately reflects the content, which focuses on integrating perception and memory research, with an emphasis on empirical foundations and computational approaches.

8/10

The event features multiple established researchers presenting empirical findings and theoretical perspectives, with a strong emphasis on integrating experimental and computational approaches. The content is scientifically rigorous, but as a community event, it includes speculative elements and open discussion, which slightly reduces the certainty of the claims.

Markers derived by PSI from the transcript: the creator did not define chapters.

• 00:00Introduction by Clementine Domine, welcoming attendees and introducing the event's theme.
• 02:00Tyler Bonnen outlines the event structure and poses the central question: how to build scalable mechanistic models of human memory.
• 05:00Akshay Jagadeesh begins his talk on visual encoding in ventral visual cortex, presenting behavioral and neural data.
• 10:00Jagadeesh discusses the texture-like representations in VTC and the divergence from human shape bias.
• 15:00Jagadeesh presents electrophysiological data from macaque IT cortex, showing similar texture and shape selectivity.
• 20:00Q&A session following Jagadeesh's talk, addressing the discrepancy between neural and behavioral responses.
• 25:00Tyler Bonnen presents his work on sequential visual processing, emphasizing the role of eye movements and time.
• 30:00Bonnen discusses lesion data from MTL patients, providing causal evidence for MTL's role in perception.
• 35:00Bonnen introduces sequential vision models that match human performance on 3D shape perception.
• 40:00Transition to the second section on modeling strategies, with discussion of computational approaches.

• CCN 2026 Community Event: Toward a unified science of perception and memory — Official event page providing details about the session and its organizers.

• Geirhos et al. (2018) - ImageNet-trained CNNs are biased towards texture — Cited in the talk as evidence that deep networks rely on texture, aligning with the discussion of texture-like representations.

• None — No discordant sources were mentioned in the video.