Skip to main content
. Author manuscript; available in PMC: 2020 Jun 1.
Published in final edited form as: Trends Cogn Sci. 2019 Apr 25;23(6):510–524. doi: 10.1016/j.tics.2019.03.006

Figure 3: Mnemonic evidence accumulation model and memory-choice cells.

Figure 3:

(A) Mnemonic evidence accumulation as a race process, with the choice made as the integrator (EV) that first reaches a preset threshold. (B) Anatomy of the PPC. Star marks the recording location for the data shown in (C-E). Adapted from [99]. (C) Example PPC neuron that increases its firing rate for a “new” decision in a graded manner modulated by confidence. Top shows the response of this neuron in individual trials, bottom the average response of this neuron. (D) Population summary of memory-sensitive neurons in the PPC reveal that some neurons signal the confidence of the decision (green), whereas others only signal the new vs. old decision (red). (E) Group PSTH of memory-choice cells in PPC, grouped according to their preferred stimulus (new or old). The preference is defined according to ground truth. Note that during errors, neurons increase their firing rate for their non-preferred stimulus, indicating that they signal choices regardless of whether they are correct or not. Adapted from [98]. (F) The balance of evidence model. Note that this model integrates the difference (new-old or old-new), which is not the same due to rectification. (G-H) Cumulative firing rate of MS neurons, shown separately for high (G) and low (H) confidence trials. The balance of evidence ΔEV scales as a function of confidence. Adapted from [58].