- [[value-based decision making]], [[subjective value]] - [[gamma oscillations]], [[gamma oscillations bridge human fMRI and money spiking activity]] # Idea By using [[intracranial EEG]] to measure local [[local field potentials LFP]] deep inside the brain (36 epileptic patients), the authors show four [[functional properties of neural value signals]]: [[properties of neural value signals - anticipatory|anticipatory]], [[properties of neural value signals - generic|generic]], [[properties of neural value signals - automatic|automatic]], [[properties of neural value signals - quadratic|quadratic]]. ## High-gamma in the OFC and HC tracked subjective value The authors' prior knowledge led them to focus on [[gamma oscillations]] (50-150 Hz) because they are thought to underlie both [[functional magnetic resonance imaging]] and spiking activity. 18 or 77 ROIs expressed subjective value (e.g., [[ventromedial prefrontal cortex|vmPFC]], [[lateral orbitofronal cortex|lOFC]], [[hippocampus]], [[parahippocampal cortex PHC]], [[anterior cingulate cortex]], [[anterior fusiform area]], [[inferior temporal cortex]], [[inferior frontal opercularis]]). All four properties were observed in the [[orbitofrontal cortex]]; all but the anticipation property was also observed in the [[hippocampus]] and [[parahippocampal cortex PHC]]. > p671. Altogether, the four functional properties were found in the OFC. All but the anticipation property were also observed in the (P)HC (Table 1). For every property, we conducted 2 × 2 comparisons (between and within the BVS ROIs) of regression estimates averaged over the relevant time windows. Compared with the OFC, the (P)HC differed only with regard to the anticipation property (t(850) = 3.37, P = 1 × 10−3 , see Table 1). Importantly, there was no significant difference between vmPFC and lOFC, which therefore shared the same four core properties. The same properties were tested in the other regions of the extended BVS: none exhibited the anticipation property but all of them shared the quadratic coding property (see Supplementary Table 8). ## Emergence of value signals in the [[brain valuation system]] The authors also examined other frequency bands (see Figure 3 results). Value signals ([[gamma oscillations]] regression coefficients) emerged in this order: - [[hippocampus]] (0.16 to 1.28s) - [[lateral orbitofronal cortex]] (0.36 to 1.00s) - [[parahippocampal cortex PHC]] (0.39 to 0.94s) - [[ventromedial prefrontal cortex]] (0.77 to 1.03s) That is, value signals emerged in the [[hippocampus]] first and in the [[ventromedial prefrontal cortex]] last, suggesting the vmPFC integrates signals. High-frequency bands contributed to likeability ratings about 10x more than lower-frequency bands. > p667. Thus, even if low-frequency activity was significantly related to subjective value, it carried redundant information relative to that extracted from high-frequency activity. ## Limitations "Pseudo-whole-brain" coverage: Limited coverage of regions (i.e., [[ventral striatum]] and [[posterior cingulate cortex]] were not measured). # References - [[Rangel 2008 A framework for studying the neurobiology of value-based decision making]] # Figures ![[Pasted image 56.png]] ![[Pasted image 57.png]]