Skip to main content


Tuesday 09 Mar 2021How can shared concept cells encode associations?

Chiara Gastaldi - EPFL Meeting ID: 985 1001 5371 Password: 289241 13:30-14:30

Experimental evidence suggests that memories of concepts such as 'the actress Jennifer Aniston' or 'the Sidney opera house' are stored in the Medium Temporal Lobe [1,2]. Each concept is represented by a sparse cell assembly ('memory engram') so that only a small fraction of neurons respond to a concept such as 'the Sidney Opera House'.   Associations between different concepts  ('Jennifer Aniston in the Sidney Opera House' ) are related to shared concept cells: assemblies representing two arbitrary concepts share less than 1% of neurons, whereas assemblies representing previously associated concepts (e.g., you have often seen Jennifer Aniston in the Sidney Opera house) share up to 4-5% of neurons in the Medial Temporal Lobe [3].  Associations between concepts can also be probed by free recall of words. Theoretical studies suggest that shared neurons between memory engrams are fundamental for the generation of sequences of words during free recall [4].

I will present a modeling study of associative memory using attractor neural networks.  I demonstrate that correlations between memory engrams induced by shared concept cells modify the memory recall process and provide a robust way to encode associations between pairs of concepts. Moreover, I will show that the free recall of a chain of concepts is possible in very sparse attractor networks only if engrams are correlated.  Finally, model predictions on the number of concepts a neuron responds to are compared with experimental data.  Our model provides a possible explanation of how association are stored in highly sparse networks (such as the Medial Temporal Lobe) and how correlated memories are recalled in the form of association chains.

[1] Quiroga, R. Quian, et al. "Invariant visual representation by single neurons in the human brain." Nature 435.7045 (2005): 1102-1107.

[2] Ison, Matias J., and Rodrigo Quian Quiroga. "Selectivity and invariance for visual object perception." Front Biosci 13 (2008): 4889-4903.

[3]  De Falco, Emanuela, et al. "Long-term coding of personal and universal associations underlying the memory web in the human brain." Nature communications 7.1 (2016): 1-11.

[4] Naim, Michelangelo, et al. "Fundamental law of memory recall." Physical Review Letters 124.1 (2020): 018101.

Add to calendar

Add to calendar (.ics)