Direct recordings catch two orbitofrontal patches taking opposite sides of a risky choice

The orbitofrontal cortex is a hard place to aim a scanner. Air-filled sinuses sit behind the eyes and distort the magnetic field, so fMRI cannot pull clean signals out of this part of the brain ^[9]. Recording from inside the skull fixes that and imposes its own limit: the volunteers are people who already have electrodes placed for another reason, in this case surgical evaluation of epilepsy ^[2].

The second design problem was engagement. Static scanner tasks tend to leave volunteers disengaged ^[10]. "They'd be falling asleep, and I thought there was no way they really cared about what they were doing," said Clara Starkweather, the lead author and a neurosurgery chief resident at UCSF ^[11]. "But then I'd see patients sitting in their beds playing Candy Crush, getting really into it" ^[12]. She built the game to reach that kind of motivation ^[17].

Edward Chang, chair of neurological surgery at UCSF and a co-senior author, located the novelty in the measurement. "We've long suspected that this region was where the brain weighs reward against risk, but we've never been able to measure it while it is happening in the human brain in real time until now," he said ^[13]. The work is published in Nature Neuroscience ^[1].

The medial patch and the lateral patch sit about two centimeters apart, and their activity runs in opposite directions ^[5][6]. An electrode placed in one is therefore two centimeters from tissue whose signal points the other way, and the study's authors offer the pair as a biomarker for calibrating targeted neuromodulation in disorders of pathological avoidance, such as OCD and depression, or compulsive risk-taking, such as addiction ^[14].

There is a second claim here, about models. Standard neurocomputational accounts build a decision by accumulating evidence continuously until a threshold is crossed ^[15]. Starkweather and co-senior author Robert Knight of UC Berkeley report modeling in which the two patches instead flip rapidly, more like an electrical switch ^[7][16].

With six patients, the weight falls on comparisons inside a person. Each trial ends in one binary choice, so a participant supplies both the risky and the safe condition ^[3]. None of the six had OCD, depression or an addiction ^[2]. No stimulation was delivered, and no accuracy figures are reported for the 500-millisecond prediction ^[18][19].