Neuroeconomic choices assume that economic decisions are centered about the activity of offer value cells in the orbitofrontal cortex (OFC), but testing this assertion has verified hard. info on PD184352 how a populace of cells contributes to a decision. Here we examined neuronal variability in the OFC of rhesus monkeys during economic decisions. Noise correlations experienced related structure but substantially lower strength compared with those typically assessed in sensory areas during perceptual decisions. In contrast, variability in the activity of individual cells was high and similar to that recorded in additional cortical areas. Simulation analyses centered on Haefner’s equation showed that noise PD184352 correlations assessed in the OFC combined with a plausible readout of offer value cells reproduced the experimental steps of CPs. In other words, the results obtained for noise correlations SIR2L4 and those obtained for CPs taken together support the hypothesis that economic decisions are primarily based on the activity of offer value cells. and were approved by the Harvard Medical School Standing Committee on Animals. In both experiments, two rhesus monkeys (1 male, 1 female) selected between juices offered in varying amounts. The two experiments differed only in the number of juices available in each session. In < 0.001]. In initial analyses, we tested the neuronal populace against a large number of variables. Procedures of variable selection identified as the three variables encoded by the neuronal populace (Padoa-Schioppa and Assad 2006). The encoding of these variables was categorical and generally consistent across time windows, indicating that neurons formed three distinct groups (Padoa-Schioppa 2013). A variable was said to explain a response if a linear regression of the response on that variable had a nonzero slope (< 0.05). For each neuron, the group was identified by the variable that provided the highest sum > 0.1; 10% of time windows excluded). Also, to minimize the effects of measurement noise, we restricted the analysis to trial types with 10 trials (4% of trials excluded). These criteria reduced the number of extreme values in the analysis, but did not substantially alter the median values obtained for and across the populace. For each neuronal response, we decided PD184352 the values of and using Deming’s regression (Glaister 2001). Simple linear regressions assume that the and PD184352 are assessed with error and the ratio between error in and error in is usually known. Errors in log() and log() were derived by propagation of uncertainty from the SE and the standard error of the standard deviation (SESD). We used an approximation of SESD (Ahn and Fessler 2003) that is usually highly accurate (within 3%) for > 10 observations: transformation. The significance of individual noise correlations was tested by computing correlations on trial-shuffled data. This method captures the range of correlations expected by chance for a pair of cells with given activity information. The trial order was randomly permuted for one of the two cells in a pair, and Pearson’s correlation was calculated on permuted data. This procedure was repeated 1,000 occasions for each pair and used to generate a confidence period. Pairs with < 0.1) did not measurably alter the effects of distance, timing, cell type, or polarity described in the main text. Similarly, repeating calculations with shorter time windows slightly reduced made it impossible to distinguish between a cell encoding with a positive slope and a cell encoding with a unfavorable slope (in both cases, the firing rate would be high/low for choices of with a positive slope (A+ cells) PD184352 and cells that encoded with a unfavorable slope (W? cells) both had polarity = +1. In contrast, A? cells and W+ cells both had polarity = ?1. The same conference held for were relabeled as encoding ((and = 229 cells). Cells with positive and unfavorable encoding were pooled (see materials and methods). Across the populace, mean(CP) ... Reconstructing choice probabilities from noise correlations. In a linear decision model, a binary decision between and is usually a linear readout of a neural populace..