Empathy, the ability to share another individual's emotional state and/or experience, has been suggested to be a source of prosocial motivation by making actions that harm others aversive. The neural underpinnings and evolution of such harm aversion remain poorly understood. Here, we characterize an animal model of harm aversion in which a rat can choose between two levers providing equal amounts of food, but one additionally delivering a footshock to a neighboring rat. We find that independently of sex and familiarity, rats reduce their usage of the preferred lever when it causes harm to a conspecific, displaying an individually varying degree of harm aversion. Prior experience with pain increases this effect. In additional experiments, we show that rats reduce the usage of the harm-inducing lever when it delivers twice, but not thrice the number of pellets than the non-preferred lever, setting boundaries on the magnitude of harm aversion. Finally, we show that pharmacological deactivation of the anterior cingulate cortex, a region we have shown to be essential for emotional contagion, reduces harm aversion, while leaving behavioral flexibility unaffected. This model of harm aversion might help shed light onto the neural basis of psychiatric disorders characterized by reduced harm aversion, including psychopathy and conduct disorders with reduced empathy, and provide an assay for the development of pharmacological treatments of such disorders. Figure 1 | Experimental Procedures. (A) Photo of the cabinet in which the experimental set-up was isolated. (B) Top view of the two-compartment set-up. (C) View of the actor's compartment through the perforated transparent divider as seen from the victim's compartment. (D) Close-up of the left wall with one of the two levers and food hoppers available to the actor. (E) Timeline of the procedures. (F) Design of exposure, baseline, shock and reversal sessions. Note that the reversal session is only present in some conditions (Table 1, 2). During exposure, the actor receives 4 shocks alone in the victim compartment. During Baseline, pressing one lever (left column) or the other (right column) leads to one pellet for the actor. The lever preferred during baseline then additionally triggers a shock to the victim during the 3 days of the Shock phase. During reversal, shocks are no longer delivered, but the non-preferred lever now leads to 3 pellets. (G) Trial structure for training, baseline, shock and reversal sessions. Grey coloring denotes events and purple, the rat's responses. (H) Simulated distribution for switching (left) and reversal indexes (right) given their respective equations and uniform distributions of preferences. Dots and lines above the distributions are the median and the 25 and 75 % percentile.
Figure 2 | Harm aversion in rats. (A) Percent choice for shock lever across sessions (Baseline, Shock1, Shock2and Shock3) and conditions (ContingentHarm, NoHarm and RandomHarm). The numbers above the graph specifie the t-statistic (t), one-tailed FDR corrected p (p1,F...