Does the medial orbitofrontal cortex have a role in social valuation?


  • *

    Present address: Laboratory of Neuropsychology, National Institute of Mental Health-National Institutes of Health, Bethesda, Maryland 20892, USA.

Dr MaryAnn P. Noonan or Jerome Sallet, as above.


It has been claimed that social behaviour changes after lesions of the ventromedial prefrontal cortex (vmPFC). However, lesions in humans are rarely restricted to a well defined cortical area. Although vmPFC lesions usually include medial orbitofrontal cortex (mOFC), they typically also affect subgenual and/or perigenual anterior cingulate cortex. The purpose of the current study is to investigate the role of mOFC in social valuation and decision-making. We tested four macaque monkeys prior to and after focal lesions of mOFC. Comparison of the animals’ pre- and postoperative performance revealed that, unlike lesions of anterior cingulate gyrus (ACCg), lesions of mOFC did not induce alterations in social valuation. MOFC lesions did, however, induce mild impairments in a probabilistic two-choice decision task, which were not seen after ACCg lesions. In summary, the double dissociation between the patterns of impairment suggest that vmPFC involvement in both decision-making and social valuation may be mediated by distinct subregions centred on mOFC and ACCg respectively.


The vmPFC region lies rostral to the ventral anterior cingulate cortex (ACC) and medial to the orbitofrontal cortex (OFC). VmPFC lesions have long been associated with alterations in social behavior and in decision-making (Bechara et al., 1997; Camille et al., 2004; Damasio, 2005; Clark et al., 2008; Rudebeck et al., 2008a). Recent reconstructions of the lesion suffered by the famous patient Phineas Gage, whose social competence changed dramatically after brain injury, have also suggested that damage to the vmPFC occurred (Damasio et al., 1994). Whilst debate has focused on the nature of the deficit the precise anatomical position of the critical lesion has received less attention. VmPFC lesions in human patients usually encompass both mOFC (Brodmann area 14) and the subgenual and/or perigenual anterior cingulate gyrus (Brodmann areas 25 and 32) while the OFC lesions in monkeys associated with changes in emotional responsiveness encompass both mOFC and lateral OFC (Izquierdo & Murray, 2004; Izquierdo et al., 2005; Machado et al., 2009). Like humans with vmPFC lesions, monkeys with OFC lesions exhibit altered emotional responsiveness (Meunier et al., 1997; Rudebeck et al., 2006) and impaired decision-making, which is usually tested in the context of visual discrimination reversal tasks (Izquierdo et al., 2004, 2005; Rudebeck & Murray, 2008).

Not only do vmPFC lesions affect social behaviour but also some imaging studies implicate the same region in social judgment. For example, we conducted a meta-analysis of functional magnetic resonance imaging (fMRI) investigations of social judgment that identified a cluster of activation in the mOFC and adjacent ACC (Fig. 1 and Supporting information, Appendix S1). Once again the meta-analysis highlighted the importance of reward-guided decision-making; activation in the same general area is found during decision-making and feedback evaluation. It is not, however, always clear whether involvement of either mOFC or adjacent ACC in social judgment can be explained by a more fundamental role in decision-making.

Figure 1.

 Meta-analysis of fMRI activations for social related processes. (A) Activations are reported on a schematic representation of a ventral view of the prefrontal cortex (adapted from Kringelbach & Rolls, 2004). (B) Activations are reported on a schematic representation of a medial view of the prefrontal cortex (adapted from Paus et al., 1996). Activations were reported in (A) the mOFC and lateral OFC as well as in (B) the adjacent anterior cingulate cortex and medial prefrontal cortex.

The orbital and medial frontal region in human and nonhuman primates is composed of multiple cytoarchitectonic areas with different anatomical connections (Petrides, 1994; Carmichael & Price, 1995a,b; Ongur et al., 2003; Haber et al., 2006). It is therefore likely that different component regions are involved in different processes. To assess the specific role of one part of the vmPFC, the mOFC, in social valuation and decision-related processes, we conducted a lesion study in monkeys. Using an identical paradigm to that used by Rudebeck et al. (2006), we tested social valuation in four macaques before and after mOFC lesions. Furthermore, utilization of the same behavioural protocol allowed us to compare the mOFC lesion results with the anterior cingulate gyrus (ACCg) lesion results obtained by Rudebeck et al. (2006).

To ascertain whether any potential impairment in social valuation was associated with impairment in fundamental aspects of reward-guided decision-making we also tested both mOFC and ACCg animals, pre- and postoperatively, on identical probabilistic two-choice decision tasks with visual stimuli. The effects of ACCg lesions on social valuation have previously been published (Rudebeck et al., 2006) but the effect of ACCg lesions on the probabilistic decision-making tasks has not been reported.

Materials and methods


Four male rhesus macaque monkeys (Macaca mulatta) aged between 7 and 10 years and weighing between 9 and 13.5 kg received mOFC lesions. All animals were maintained on a 12-h light–dark cycle and had 24-h ad lib access to water, apart from when they were testing. All experiments were conducted in accordance with the United Kingdom Scientific Procedures Act (1986).

Surgery and histology

The following section summarizes the details of the surgery, anesthesia and histological protocols for the mOFC-lesioned animals. Procedures specific to the lesions made in the comparison groups in orbital and ventrolateral prefrontal cortex (PFv+o) and anterior cingulalte gyrus (ACCg) have been published previously (Rudebeck et al., 2006). In the current study, at least 12 h before surgery macaques were treated with an antibiotic (8.75 mg/kg amoxicillin, i.m.) and a steroidal anti-inflammatory (20 mg/kg methylprednisolone, i.m.) to reduce the risk of postoperative infection, oedema and inflammation. Additional supplements of steroids were given at 4- to 6-h intervals during surgery. On the morning of surgery, animals were sedated with ketamine (10 mg/kg, i.m.) and xylazine (0.5 mg/kg, i.m.) and given injections of atropine (0.05 mg/kg), an opioid (0.01 mg/kg buprenorphine) and a nonsteriodal anti-inflammatory (0.2 mg/kg meloxicam) to reduce secretions and provide analgesia, respectively. They were also treated with an H2 receptor antagonist (1 mg/kg ranitidine) to protect against gastric ulceration, which might otherwise have occurred as a result of administering both steroidal and nonsteroidal anti-inflammatory treatments. Macaques were then moved to the operating theatre where they were intubated, switched onto isoflurane anesthesia (1–2%, to effect, in 100% oxygen), and placed in a head holder. The head was shaved and cleaned using antimicrobial scrub and alcohol. A midline incision was made, the tissue retracted in anatomical layers, and a bilateral bone flap removed. All lesions were made by aspiration with a fine-gauge sucker. Throughout the surgery, heart rate, respiration rate, blood pressure, expired CO2 and body temperature were continuously monitored. At the completion of the lesion, the wound was closed in anatomical layers. Nonsteroidal anti-inflammatory analgesic (0.2 mg/kg meloxicam, orally) and antibiotic (8.75 mg/kg amoxicillin, orally) treatment was administered for 5 days postoperatively. All surgery was carried out under sterile conditions with the aid of a binocular microscope. The wound was closed in anatomical layers. At least 2 weeks were allowed for recovery before testing resumed.

When the animals had completed their testing they were anesthetized with sodium pentobarbitone and perfused with 90% saline and 10% formalin. The brains were then removed and placed in 10% sucrose formalin until they sank. The brains were blocked in the coronal plane at the level of the most medial part of the central sulcus. Each brain was cut in 50-μm coronal sections. Every tenth section was retained for analysis and stained with Cresyl violet.

Experiment 1


All training and testing was conducted while the animals were in a transport cage inside a modified Wisconsin General Testing Apparatus (WGTA; Fig. 1A). A Plexiglas box measuring 70 × 11 × 11 cm with a hinged back was fixed to the WGTA 20 cm in front of the transport cage. In addition behind the box, 50 cm from the front of the transport cage, was a PC monitor for presenting visual stimuli. On each trial, stimuli could be presented either in the box or on the PC monitor. The stimuli presented in the box could be one of the following: 20 neutral control ‘junk’ objects and two fear-inducing stimuli (a static rubber snake or a moving wooden snake). Stimuli presented on the screen were video clips 30 s in length and were one of two human stimuli (two unknown staring human faces), one of five social stimuli [a large (11 kg) monkey staring, a monkey (8 kg) exhibiting affiliative behaviours (lip-smacking), a monkey (9 kg) inspecting a transport cage or a monkey (5 kg) with food, and a female macaque (5 g) with prominent perineal swelling] or a moving, neutral control stimulus (a moving randomly changing coloured object; Fig. 3B). The video clips were chosen because it made it possible to compare the effects of mOFC lesions with the effects of ACCg lesions; the stimuli had previously been used in an investigation of the effects of ACCg lesions (Rudebeck et al., 2006). The social stimuli were chosen because they were expected to elicit varying degrees of interest from control monkeys (and indeed this proved to be the case as explained below). Video stimuli were clips taken from longer videos of other monkeys in the colony recorded while they were in either transport cages or primate chairs. At the time of testing all the monkeys in the video social stimuli were novel to the subjects. All videos were taken using a Panasonic EZ35 mini-DV camera and edited using Adobe Premier Pro 1.5 software. Videos were played using Windows media player version 9.0.

Figure 3.

 WGTA testing apparatus and representations of stimuli. (A) Monkeys sat in a transport cage facing the perspex box and PC monitor. On each trial the moveable screen was raised and animals could take the valued food item (peanut or quarter date) placed at the far edge of the perspex box. After 30 s the screen was lowered regardless of whether the animal had taken the food or not. Represented below are the stimuli (B) used including snakes and individual frames of the five videos of macaque social stimuli. From left to right: monkey inspecting cage, monkey with food, monkey making affiliative gestures, female monkey perineum and staring monkey. Additional stimuli used included the human social stimuli and neutral stimuli.

A camera mounted on top of the PC monitor recorded each monkey’s behaviour. The camera faced the whole cage and allowed monkeys’ latencies to take food rewards placed at the back of the Perspex box to be measured and general behaviour and facial expressions to be recorded for later analysis.


Four macaque monkeys (Macaca mulatta) were tested on a social valuation task (Rudebeck et al., 2006) before and after mOFC lesions. Briefly, animals were tested in the WGTA (Fig. 3A) and on every trial the monkey retrieved a small food item that was placed in a fixed central position on the top of a transparent plastic box. Two different emotive toy snakes (static and moving) were used to investigate fearfulness (experiment 1a). The five short films of other macaques (detailed above) were used to investigate social valuation in experiment 1b. Responsiveness to videos of humans staring was also assessed in experiment 1c. Finally, responsiveness to neutral control objects was also assessed in order to provide a baseline against which to compare any changes in fearfulness and social valuation (experiment 1d). On each trial, stimuli were placed in the Perspex box or displayed on a screen behind the box. The animal had 30 s to retrieve the food item or else an opaque moveable screen was lowered in between the animal and the box for the duration of a 30-s intertrial interval. The latency to reach for the piece of food indexed the macaques’ assessments of the value of obtaining additional information about the stimulus before reaching, and reflected their relative valuation of the stimulus in contrast to the incentive value of the food. On each day, animals were exposed to ten different stimuli of possible social or emotional importance and 20 neutral objects. The test was repeated over four sessions (with a day of rest in between sessions) and the median reaching latency for each stimulus per animal was calculated. Each stimulus either in the box or on the screen was presented once per day. Objects in the box and images on the screen were presented in a pseudorandom order. The constraints enforced on order were that neutral object trials always followed trials in which potentially fear-inducing or social stimuli (snake or monitor stimuli) were presented. In experiment 1 two animals (mO1 and mO2) had acted as unoperated controls in a previous experiment (Rudebeck et al., 2006). The other two animals (mO3 and mO4) were tested only in this study pre- and postoperatively. The data from all four animals were considered together because there was no discernable difference between animals’ performances in relation to the time of testing.

Habituation and reward preference test

Animals were first habituated to the testing environment and then trained to take food from the top of the Perspex box while it was empty. The food reward was located at the centre of the back edge of the box nearest the PC monitor so that during the actual test the animal would have to reach over anything in the box or as close as possible to the monitor. A small mark on the back of the box meant that the food reward was always put in the same place. Each trial began with the movable screen being raised. The monkey then had 30 s to retrieve the food reward located on the box. The screen stayed up regardless of whether or not the monkey took food reward within the 30 s. At the end of the trial the screen was lowered for 30 s before the next trial. During this period the experimenter could change the object in the box or image on the monitor and replace the food item. Before the screen was raised for the next trial a curtain that obstructed the animal’s view of the experimenter was fixed to the back of the WGTA. The curtain was used to ensure that monkeys could not see the experimenter during the trial as the presence of a human could have affected later trials involving human or monkey stimuli presented on the screen.

For the test to assess emotional and social value of the different stimuli the food rewards had to be motivationally significant. We therefore needed to find a food highly valued by each individual animal. All animals were initially trained to take a single peanut food reward. A food reward was judged as motivationally significant if the animal took the food item from the back of the box in < 5 s for 20 consecutive trials. Animals who did not reach this criterion with peanut food reward were trained to criterion with a quarter piece of date. This food object was then used throughout the rest of training and testing. Over a further 3 days they were then trained to take their preferred food reward from the top of the box while any one of nine novel ‘junk’ objects were presented inside a moving changing coloured object presented on the computer screen positioned behind the box. Objects were presented in sets of five per day with each object being presented twice (10 trials). These ‘junk’ objects were not used subsequently during testing and instead further sets of novel junk objects were used in the interleaved control trials in the tests of emotion and social behaviour.

Data acquisition

Each trial was recorded on VHS video and analyzed independently by two raters (M.P.N. and J.S.) Reaching latencies were measured from the beginning of the trial, as defined by the raising of the screen, to the time the animals first grasped the piece of food. Despite high inter-rater reliability (Pearson correlation r = 0.986), all trials in which there was a discrepancy of > 40 ms between the two raters’ scores were re-evaluated. Forty-six out of 960 trials were identified in this manner and re-analysed by both raters.

Data analysis

The start of each trial was initiated when the screen was raised above a fixed point marked on the side of the cage at approximately the same height as the top of the Perspex box. For the reaching latency measurement, the response was considered finished at the point just before the animal moved the food object from its initial position. If the animals did not retrieve the food reward within the 30 s, a score of 30 s was given. Latency measurements were scored to the nearest frame (0.04 s).

The data from experiment 1a was subjected to a three-way repeated-measures anova with factors of surgery (two levels: pre- and postoperative), session (four levels: 1–4 days), and stimuli (two levels: moving and static snake). The first two factors were also used in the three-way repeated-measures anovas used to analyze the data from all the other experiments but then the third factor either reflected the five levels of social stimuli (monkey inspecting cage, monkey with food, monkey making affiliative gestures, female monkey perineum and staring monkey) in experiment 1b, the two different human video stimuli (experiment 1c), or the two different classes of neutral stimuli (moving or static objects). Reaching latencies were log-transformed when necessary in order to minimize the impact of positive skewing and to reduce between group differences in reaching-latency variance.

In addition to measuring reaching latencies to the food, two experimenters (J.S. and M.P.N.) scored each animal’s behaviour in response to each stimulus using an adapted form of the checklist employed by Aggleton & Passingham (1981) (Izquierdo & Murray, 2004; Izquierdo et al., 2005; Rudebeck et al., 2006). The behavioural responses were categorized into affiliative behaviour (lip-smacking) and aggressive or conflict behaviour (ears flat, open-mouth threat, piloerection and cage shaking). Each instance of a behaviour in each relevant behavioural category during the 30-s trial period was recorded and their mean frequency was compared pre- and postoperatively. Because the stimuli in the present experiment, as in the study of Rudebeck et al. (2006), were never used to directly threaten the animal they were far less effective in eliciting strong behavioural responses than those used by Aggleton and Passingham. A three-way within-subjects anova compared the responses of the animals pre- and postoperatively (lesion) with respect to the two behavioural categories (social or affliative, and aggressive or conflict) to the five social stimuli (stimuli: staring human, female monkey perineum, staring monkey, moving snake and moving pattern).

Subsequent analyses compared the effects of mOFC lesions with those induced by lesions to other regions of the frontal lobe. Previously collected data from animals with ACCg lesions were compared to the mOFC postoperative testing sessions. Four independent two-way repeated-measures anovas mirrored the analyses described above. Emotional stimuli were compared in a three-way anova of stimulus, session and the between-subjects factor of lesion position (mOFC or ACCg). Social stimulus effects were compared in a three-way anova of social stimuli, session and the between-subjects factor of lesion position. Responses to human video stimuli were compared in a three-way anova of social human stimuli, session and the between-subjects factor of lesion position. Finally, responses to neutral stimuli were compared in a three-way anova of neutral stimuli, session and the between-subjects factor of lesion position.

Experiment 2


Each monkey sat in a testing room, unrestrained, in a wheeled transport cage placed 20 cm from a touch-sensitive monitor (38 cm wide × 28 cm high) on which pairs of visual stimuli could be presented (eight-bit colour clipart bitmap images, 128 × 128 pixels) and responses recorded. Rewards (190-mg Noyes pellets) were delivered from a dispenser (MED Associates, St Albans, Vermont) into a food well immediately to the right of the touch screen. A large metal food box, situated to the left below the touch screen, contained each individual’s daily food allowance (given in addition to the reward pellets) consisting of proprietary monkey food, fruit, peanuts and seeds, delivered immediately after testing each day. This was supplemented by a forage mix of seeds and grains given ∼6 h prior to testing in the home cage. Stimulus presentation, experimental contingencies, reward delivery and food box opening was controlled by a computer using in-house software.


The mOFC animals were tested pre- and postoperatively on a simple two-choice task. Before the start of testing, all macaques had received extensive training with touch screens and knew that touching a stimulus on the screen could lead to food reward. Each day, macaques were presented with two novel stimuli on the touch screen at the same time in a left/right configuration. Each stimulus’ side of presentation varied from trial to trial. On each trial, selecting one stimulus caused the other to extinguish and reward to be delivered according to the reward schedule. Auditory tones were used to cue the animal to the presentation of the stimuli, to the selection of a stimulus and to the potential delivery of a reward. Each stimulus was associated with a different outcome probability, one stimulus always being rewarded more than the other. At the start of testing, each stimulus was randomly assigned one of two reward probabilities (Fig. 6A). The ratios of reward associated with the two stimuli were either 75 : 25 (in other words one stimulus had a 0.75 probability of reward while the other had a 0.25 probability of reward) or 50 : 18. Each schedule was performed twice and in an interleaved manner. Monkeys’ touches registered their stimulus selections. Upon a decision being made rewards were delivered according to a specific schedule (75 : 25 and 50 : 18) with a fixed probability with a reward matching contingency in place (Herrnstein, 1997; Sugrue et al., 2004; Kennerley et al., 2006; Rudebeck et al., 2008b). This meant that rewards once allocated to a stimulus remained available until that stimulus was chosen. Further details can be found in Rudebeck et al. (2008b).

Figure 6.

 Schematic representation of the decision-making task and results. (A) On each trial, monkeys were presented with two stimuli each associated with a different and fixed probability of receiving a reward with a matching contingency in place. Gray, blue and red circles represent different 250 ms tones which served as cues to the animal to the presentation of the stimuli, to the selection of a stimulus and to the potential delivery of a reward. The animals experienced two reward-matching schedules in which the ratios of reward associated with the two options were 50 : 18 and 75 : 25. (B and C) Performance in experiment 3. Mean number of trials, across the two testing sessions administered at each reward ratio, taken by each macaque to achieve the ratio of responses associated with 97% of the maximum reward rate (ropt). The unfilled bars represent performance of the mOFC group prior to lesion and the CON group (B and C respectively), whereas grey represent mOFC and ACCg groups (B and C respectively). The performance of individual macaques, within each group, is represented by distinct symbols.

The performance of each macaque was assessed by calculating the number of trials to reach the ratio of responses (high-probability stimulus choices to low-probability stimulus choices) that yielded 97% of the maximum rate of reward (ropt; Kennerley et al., 2006). In brief, ropt was calculated by plotting the expected value (EV) given each rate, r, of high stimulus choice, EV(r), for the different probability ratios (75 : 25, 50 : 18), where p and q represent the probability of reward associated with the high and low stimuli (Equation 1, below); ropt was then determined for each probability ratio by taking the maximum point on each of the curves plotted. The number of trials that macaques took to achieve 97% of the ropt was determined using a 20-trial moving window (−10/+10) of the subjects’ choices for the high reward-probability stimulus.


If criterion was not reached by the end of the 100-trial session a score of 100 was allocated to that animal on that session. The results were subjected to a repeated-measures anova of lesion (pre- and postoperative) × reward ratio (50 : 18 and 75 : 25) × session.

The ACCg animals were tested and analyzed in a similar way although they were compared to a group of unoperated control animals (N = 6) in a three-way anova with a two-level factor of reward ratio (50 : 18 and 75 : 25) × two-level factor of session × the between-subjects factor of group (ACCg; unoperated controls). It is difficult to make direct comparisons between the postoperative performances of the mOFC and ACCg animals as three of the unoperated control animals that were used in the ACCg experiment were subsequently tested as part of the mOFC group. This means that at the final mOFC postoperative test these animals had been tested three times on this paradigm whereas the postoperative ACCg animals had never previously been tested on a reward-matching task although they had had experience of other reward-guided visual discrimination tasks.



As can be observed in Fig. 2, the mOFC lesions were made as intended. These lesions included mainly Walker area 14. For full details of the ACCg lesion please refer to Rudebeck et al. (2006). In brief, however, the ACCg lesions were largely confined to area 32 and the anterior ventral tiers of area 24.

Figure 2.

 Diagram of mOFC intended (left) and actual (right) lesions for the four experimental animals (mO1, mO2, mO3 and mO4).

Experiment 1

Reaching latencies

MOFC lesions produced no significant effects on reaching latencies for any fear-inducing stimuli in experiment 1a (Fig. 4A; either main effect of mOFC lesion; F1,3 = 0.014, = 0.912, or interaction of lesion with stimulus type; F1,3 = 0.045, P = 0.845). There was, however, a main effect of stimulus type (F1,3 = 27.84, P = 0.013), with the animals being slower to reach for the moving snake than the rubber snake.

Figure 4.

 Response to mild fear and social stimuli in experiment 1. (A) Response to mild fear and social stimuli in experiment 1. Median latency to retrieve food in the presence of a mild fear-inducing stimulus [(i), static or moving snake], social stimuli, either (ii) unknown macaques or (iii) human actors, and (iv) neutral control stimuli. Symbols indicate scores for each individual. No changes in the mean latency to retrieve the food placed in front of any different stimulus types were seem after mOFC lesion. (B) The effect of mOFC lesions on measures of affiliative behaviour [(i); lip-smacking] and combined measures of commonly observed aggression and conflict behaviours [(ii); ear flattening, open mouth threat, cage shaking, and piloerection]. Symbols indicate scores for each individual. Overall no differences in the number of affiliative or aggressive responses were observed before and after mOFC lesion.

There was also no effect of lesion (F1,3 = 0.41, = 0.568) or interaction of lesion with stimulus type (F4,12 = 1.30, P = 0.327) for reaching times to the social stimuli in experiment 1b. However, a linear main effect of social stimulus type was revealed (F1,3 = 3.48, P = 0.040), suggesting that animals, regardless of the presence of an mOFC lesion, agree as to which images of other macaques are the most interesting (Deaner et al., 2005, Rudebeck et al., 2006, Klein et al., 2008, 2009). All animals were slower to retrieve food in the presence of a large staring male, a female macaque with visible sexual perineal swellings or a midsized macaque making affiliative lip-smacking gestures than was the case with the other less socially salient images. There was also no effect of mOFC lesion on reaching latencies for the social human stimuli in experiment 1c (F1,3 = 2.53, = 0.210) or interaction between the mOFC lesion and human stimulus type (F1,3 = 0.91, = 0.410).

Finally there was no effect of mOFC lesion on reaching latency in the presence of neutral stimuli (main effect: F1,3 = 1.25, P = 0.345; interaction of mOFC lesion and neutral stimulus category: F1,3 = 2.332, P = 0.0.224). There was, however, a three-way interaction found between lesion, neutral stimuli and session (F3,9 = 4.21, P = 0.041) and a main effect of neutral stimuli (F1,3 = 22.56, = 0.018). Inspection of the data suggests that this three-way interaction can be attributed to longer reaching latencies, in the first testing session pre-operatively, towards moving stimuli only. The main effect was due to longer reaching latencies towards the moving stimuli regardless of the presence of lesion (paired samples t-test: preoperative, t3 = −3.06, P = 0.055; postoperative, t= −3.15, P = 0.051).

To note, we observed effects of habituation in the responses to all four stimulus types. One-way anovas of session (four levels: four testing days) and fear stimulus (two levels: moving and static snake) revealed a near main effect of session (F3,9 = 4.77, P = 0.068), which individual one-way anovas attributed to habituation to the static snake only (F3,9 = 4.89, P = 0.028); the moving snake did not elicit habituation effects over testing session (F3,9 = 0.77, = 0.536). Analyses of the other stimulus types revealed a main effect of session for the social monkey stimuli (F3,9 = 11.92, P = 0.005) and social human stimuli (F3,9 = 11.53, = 0.002). Effects of session on the neutral stimuli on tended to significance (F3,9 = 4.19, P = 0.091).

Behavioural responses to stimuli

Not only did the mOFC lesion not alter monkeys’ reaching latencies to the various categories of stimuli but it did not greatly alter any other measure of their social interaction during the test (Fig. 4B). Analysis of the frequency of certain social behaviours revealed very few significant effects. MOFC lesions produced no differences in the frequency with which aggressive or affiliative behaviors were displayed. There was no effect of lesion (F1,3 = 2.99, = 0.182), Behavioural category (F1,3 = 0.71, P = 0.461) or social stimuli (F4,12 = 0.77, = 0.507). There was, however, a significant interaction of lesion with stimuli (F4,12 = 5.67, = 0.008) which appears to be as a result of fewer behavioural responses elicited towards the human staring stimuli after mOFC lesions (two-tailed paired-samples t-test: t= 2.45, = 0.092 and t= 5.00, = 0.015; affiliative and aggressive respectively). These results imply that mOFC-lesioned animals did not greatly change their overt behaviour to macaque social or emotional stimuli even if there was a reduction in the responsiveness towards videos of people.

Comparison of mOFC and ACC lesions

An additional analysis directly compared the effect of mOFC and ACCg lesions on the same social valuation test (Rudebeck et al., 2006). Figure 5A illustrates the intended lesion location for the mOFC and ACCg animals.

Figure 5.

 A comparison of the effects of mOFC and ACCg lesions on behaviour elicited by mild fear-inducing and social stimuli. (A) Intended mOFC and ACCg lesion locations. (B) MOFC and ACCg lesion groups’ reaching latencies in the presence of (i) mild fear-inducing stimuli and (ii) the macaque social stimuli. There were no group differences in reaching latencies towards fear-inducing stimuli, but ACCg lesion animals had faster reaching latencies in the presence of socially relevant stimuli than did mOFC-lesioned animals.

In a comparison of the two groups’ responses to the fear-inducing stimuli no differences between the effects of the two lesions were seen. Specifically, there were no interactions involving group (fear stimuli × group, F1,5 = 1.04, = 0.355, fear stimuli × session × group, F3,15 = 0.72, P = 0.513) nor main effects of group (F1,5 = 4.38, P = 0.090). The only main effect of interest related to the identity of the fear stimuli (F1,5 = 11.70, = 0.019). This implies neither the mOFC nor the ACCg have fundamentally critical roles in guiding this type of behaviour.

In contrast, a comparison of group responses towards the social stimuli (pictures of other monkeys) revealed that the ACCg was the critical region for social valuation (Fig. 3D). There was a significant linear main effect of the identity of the social monkey stimuli on responsiveness (F1,7 = 7.37, P = 0.030), confirming that the monkeys whose behaviour was investigated concurred with one another in their valuations of the videos of other monkeys. There was a significant interaction of social monkey stimulus, session and group (ACCg vs. mOFC) on the log-transformed reaching latencies (F12,60 = 2.45, P = 0.016), in addition to a significant main effect of the identity of the social monkey stimuli (F4,20 = 3.83, = 0.029).

An analysis that compared the two lesion groups’ responses to the human stimuli found no significant group differences (F1,5 = 1.54, P = 0.269) or interaction with the stimulus identity (F1,5 = 0.058, P = 0.819). Similarly, there were no significant group differences in an analysis of the neutral stimuli (F1,5 = 0.36, P = 0.573) or interactions between group and stimulus identity (F1,5 = 2.10, = 0.207). A main effect of neutral stimuli was noted (F1,5 = 13.78, P = 0.014); it was a result of longer reaching latencies towards the moving pattern stimuli that the neutral static objects (paired-samples t-test: preoperative, t3, = −3.15, P = 0.051; postoperative, t= −3.06, = 0.055).

Not only did Rudebeck et al. (2006) demonstrate that performance in the social valuation task was altered by ACCg lesions but they also reported that lesions of ventrolateral and lateral orbital prefrontal cortex (PFv+o) did not alter monkeys’ reaching latencies in response to social stimuli but that they did affect responsiveness to fear-inducing stimuli (Rudebeck et al., 2006). Interestingly, when the mOFC animals’ reaching latencies in the context of fear-inducing stimuli are compared to those of animals with PFv+o lesions, it can be seen that the PFv+o-lesioned animals were significantly faster to reach for food placed over emotive stimuli than were animals with mOFC lesions (supporting Fig. S1; fear × group interaction, F1,5 = 9.29, P = 0.028; and main effect of group, F1,5 = 9.59, = 0.027), suggesting that they were less fearful. In contrast, no differences were found between the two lesion groups in their responses to the social stimuli (social monkey stimuli × session × group; F12,60 = 1.30, P = 0.031).

Experiment 2

Although the mOFC plays no fundamental role in social valuation or emotional responsiveness it was implicated in the two-choice decision-making task (Fig. 6A). Analysis of the data shown in Fig. 6B reveal a main effect of mOFC lesion (F1,3 = 44.17, P = 0.007). In contrast, when the ACCg-lesioned animals were compared to their matched controls (Fig. 6C) no lesion deficit was apparent; there was neither a main effect of the lesion (F1,7 = 2.00, P = 0.201) nor any interaction between the effect of the lesion and the particular type of decision-making task (F1,7 = 0.02, = 0.889). This suggests that mOFC may have the more important role in decision-making.


The study examined the effects of mOFC lesions (centred on area 14) on social and emotional valuation and reinforcement-guided stimulus selection, and then compared them with that of ACCg lesions (centred on areas 24a, b and 32). Contrary to our predictions, mOFC lesions caused no impairments in social valuation or emotional responsiveness. The animals were equally reluctant to reach for food in the presence of fear-inducing stimuli both before and after mOFC lesions. Similarly, there was no change in animals’ assessments of how interesting each social stimulus was as indexed by reaching latencies before and after their lesion, nor did we observe an alteration in other aspects of behaviour in the context of the social stimuli.

The lack of change in social valuation or fearfulness in the mOFC lesion group cannot be attributed to a lack of sensitivity in the task; the task was sensitive to altered social valuation in animals with ACCg lesions and to altered emotional responsiveness in animals with PFv+o lesions (Rudebeck et al., 2006). A formal comparison demonstrated that the ACCg lesion animals were significantly less interested in the social stimuli than were the animals with mOFC lesions. Moreover, the null effect of the mOFC lesion on the social and fear tasks cannot be attributed to some deficiency in the surgery; the mOFC-lesioned animals, but not the ACCg-lesioned animals, were impaired in the decision-making task. Experiment 2 showed that mOFC lesions disrupt the ability to choose the better value stimulus option. There is also evidence that the mOFC decision-making deficit becomes more severe when animals choose between more than two different stimuli (Noonan et al., 2010). In summary, there was evidence for a double dissociation between the effects of ACCg and mOFC lesions on social valuation and reward-guided decision-making.

The absence of impairment in social and emotional processing after mOFC lesions initially appears to conflict with the prevalent view that vmPFC lesions cause disinhibited and socially inappropriate behaviour (Damasio, 1994; Rolls et al., 1994; Anderson et al., 2006; Blake & Grafman, 2006). VmPFC and OFC lesions have been implicated in a variety of emotional and decision-making deficits (Bechara et al., 2000; Fellows, 2007; Clark et al., 2008; Heberlein et al., 2008). However, the lesions in patients are often a result of stroke or head trauma and as a result the damage is often not restricted to one cortical area. In conjunction with the previous study reported by Rudebeck et al. (2006), the present results suggest that it is not damage to the mOFC in patients with vmPFC lesions which causes alterations in social behaviour but rather damage to the subgenual and perigenual cingulate cortex and possibly to the medial frontopolar cortex (Bechara et al., 1997; Camille et al., 2004). Furthermore, loss of the white matter tracks underlying damaged cortex may contribute to impairments (Philippi et al., 2009).

It remains a possibility that while mOFC is not essential for simple social valuation it is important for more complex judgments involving regret or guilt (Saver & Damasio, 1991; Camille et al., 2004; Koenigs & Tranel, 2007; Koenigs et al., 2007; Krajbich et al., 2009). However, judgments about regret do not just reflect broader social considerations but they also require consideration of counterfactual outcomes that are then compared with actual outcomes. It has recently become clear that information about counterfactual outcomes is represented in parts of frontopolar cortex (Boorman et al., 2009) that may also be damaged in patients with vmPFC lesions. Judgments about guilt may also require knowledge of one’s own or another person’s intentions and therefore depend on paracingulate areas implicated in theory of mind (Amodio & Frith, 2006; Frith & Frith, 2006; Behrens et al., 2008; Hampton et al., 2008).

It is also possible that the mOFC is more important in complex social situations in which choices have to be made between many different possible courses of action. We have found evidence that the macaque mOFC is especially important when decisions have to be made between multiple options that are all associated with different levels of reward (Noonan et al., 2010). This suggests that mOFC might be more important in complex social decision-making settings that require consideration of the benefits of several different possible choices.

Previous investigations of large OFC lesions have reported reduced fearfulness and increased aggressiveness (Izquierdo et al., 2005; Machado & Bachevalier, 2006, 2008). The work of Machado & Bachevalier (2006) suggests that the lesion in the lateral part of the OFC (area 11 and 13) may have been critical for causing these deficits. Rudebeck et al. (2006) previously showed that animals with PFv+o lesions, which included lateral OFC, were significantly less fearful than control animals and animals with ACCg lesions. Here we show that the same animals with PFv+o lesions were also significantly less fearful than animals with mOFC lesions animals; PFv+o lesion animals were significantly faster to reach for food placed over emotive stimuli than were animals with mOFC lesions. Neither mOFC lesions in the present study nor lesions that included lateral OFC (Rudebeck et al., 2006) altered social valuation. Furthermore, mOFC-lesioned animals did not display any other changes in their general behaviour emotional responsiveness to the various stimuli (Fig. 4B).

The lack of importance of the mOFC in the analysis of social stimuli can perhaps be understood in the context of its anatomical connections. Indeed, the mOFC does not receive direct inputs from temporal areas involved in processing macaque vocalizations (i.e. temporal auditory areas; see Ghazanfar et al., 2005 and Romanski & Averbeck, 2009) or faces (areas TE and TEO; see Webster et al., 1994 and Carmichael & Price, 1995a). FMRI studies conducted with macaques have demonstrated lateral OFC responsiveness to images of faces (Tsao et al., 2008) while the ACCg is particularly responsive to the vocalizations of conspecifics (Gil-da-Costa et al., 2004).

Valuation of social information is also an important determinant of activation in the human ACC. Behrens and colleagues (Behrens et al., 2008, 2009) found that ACCg activation to the delivery of feedback after decision-making increased in proportion to the importance of the feedback for finding out about another person. The subjects studied by Behrens and colleagues played an interactive decision-making game with another player. Feedback was more important for finding out about the other player in phases of the game when the other player’s behaviour was changing more rapidly; it was at these points in the game that outcome-related ACCg activity was highest. Predictions and prediction errors concerning the other player’s intentions were associated with changes in activation in paracingulate cortex. Such information about the other player was then used, in conjunction with the subject’s own choice–reward history, to estimate the probability of obtaining a reward on each trial of the game and this estimate was associated with mOFC activation. The dissociation between ACCg activation during the valuation of social information and mOFC activation in relation to reward-guided decision-making mirrors the dissociation between the impairments found after lesions to the two areas in the current experiment. The studies suggest that while mOFC may be active in social decision-making contexts (Fig. 1) its activation reflects expectations about the rewards or other benefits that the subjects hopes to obtain from the decisions that are made.

Because the mOFC is active in social situations, albeit in a manner that reflects the benefits for the subject that might be obtained from the social situation (Behrens et al., 2008, 2009), it is to be expected that lesions that affect the mOFC are likely to alter the decisions that people and animals make in complex social situations. There is certainly evidence for changes in behaviour in monkeys with OFC lesions in naturalistic and complex social situations (Machado & Bachevalier, 2006, 2008). Such changes may partly reflect the consequences of more primary alterations in animals’ fearfulness and aggression that occur as a result of damage to lateral parts of the OFC (Rudebeck et al., 2006). In addition, alterations in behaviour in complex social situations after OFC lesions may partly reflect the role that the mOFC has in making reward-based decisions in situations where there are many possible choices (Noonan et al., 2010).

In summary, the mOFC appeared to have no critical role in social valuation or in mediating emotional responsiveness. Instead the mOFC seems more involved in comparing the values of choices as illustrated by the decision-making deficit in experiment 3. VmPFC lesion patients with socially inappropriate behaviour may have damage that extends into the ACCg region, which appears to be far more critical for social valuation. The inappropriate behaviour exhibited by vmPFC patients may be a result of an inability to evaluate the outcome of their socially orientated actions or the potential reaction of the other person.


This research was supported by MRC and Wellcome Trust.


anterior cingulate cortex


anterior cingulate gyrus


functional magnetic resonance imaging


medial OFC


orbitofrontal cortex


orbital and ventrolateral prefrontal cortex


maximum rate of reward


ventromedial prefrontal cortex or cortical


Wisconsin General Testing Apparatus