Brain Inspired BI 202 Eli Sennesh: Divide-and-Conquer to Predict
4 snips
Jan 3, 2025 In this engaging discussion, Eli Sennesh, a postdoctoral researcher at Vanderbilt University, sheds light on predictive coding and its implications for understanding brain functions. He navigates the intriguing concept of 'divide-and-conquer predictive coding' and its experimental applications. The conversation also touches on the relationship between neuroscience and AI, emphasizing the need for biologically plausible computational models. They explore the complexities of decision-making, consciousness, and the humor in our perceptions of task difficulty, offering a delightful blend of research insights and personal anecdotes.
AI Snips
Chapters
Books
Transcript
Episode notes
Bridging Theory and Experiment
- Eli Sennesh highlights the challenge of connecting theory to experiments in neuroscience.
- He emphasizes that null results are expected, but linking theory and experimental design is crucial.
Oddball Paradigm Limitations
- Standard neuroscience paradigms, like the oddball paradigm, conflate surprise with other factors like visual change.
- Eli Sennesh questions how these paradigms isolate surprise processing.
Model Comparison
- Eli Sennesh discusses the challenge of differentiating between competing computational models that fit the same behavioral data.
- He questions whether researchers sometimes "pretend" their model is the correct one without rigorous comparison.
