Fig. 1
From: Neuronal allocation and sparse coding of episodic memories in the human hippocampus
Empirical quantile-quantile (Q-Q) plots of the target- vs. foil-by-neuron normalized spike-count distributions at retrieval. The target-by-neuron response distribution (y-axis) vs. the foil-by-neuron response distribution (x-axis) was compared separately for the hippocampus (a, c) and amygdala (b, d). The top panels (a, b) plot the distributions of 100% of the normalized spike counts from each neuron in response to each retrieval trial (item presentation), across all patients (N = 79) and sessions (N = 55), in the hippocampus (a: 72,538 recordings from 736 neurons) and amygdala (b: 99,944 recordings from 1,038 neurons). Densely grouped points (dark gray) represent thousands of recordings, and less densely grouped points (light gray) represent relatively few recordings. In panel (a), the nonlinear upward deflection toward the y-axis reflected that the hippocampal target-by-neuron response distribution had a different shape than the foil-by-neuron response distribution, and specifically, a more positive skew. The deflection in (a) is predicted by a sparse coding account, which further predicts that the deflection reflects the strong responses of a small percentage of neurons in response to target items. A less pronounced deflection in the opposite direction was visible in the amygdala (b). After removing the top 0.25% of the target and foil spike counts for the hippocampus (c) and the amygdala (d), the deflections in (a) and (b) were no longer apparent. The skewness values of the two normalized spike count distributions (targets vs. foils) were compared to determine if the theoretically predicted deflection in panel (a) was statistically reliable using bootstrap tests (B = 10,000, p < 0.05, one-tailed). The difference in skewness between the target- vs. foil-response distributions was significantly different between the brain regions (hippocampus vs. amygdala). In the hippocampus only, skewness was greater for targets than foils, but this difference did not reach significance (p = 0.070). Refer to Table 1 for detailed statistical reporting. *p < 0.05, ~p < 0.10, one-tailed.
