Being able to attenuate fear by using information about the long-term reward associated with a scary prospect is a key aspect of emotional intelligence. Building up on our behavioural results (Neural mechanisms underlying the control of emotion -Part 1: Behavioural dataset - see 'Related Resources' section below), our goal here was to elucidate the neural correlates of this emotion regulation skill. Emotion regulation has often been studied in the laboratory with re-appraisal paradigm, where participants are asked to re-appraise stimuli that have been paired with aversive outcomes in a positive manner. For example, participants may be asked to use the color of a fear-conditioned stimulus to trigger an image of a relaxing scene. However, because the re-appraisal paradigm is open-ended, we know little about how re-appraisal instructions were implemented at a mechanistic level. Here we used joint appetitive-aversive classical conditioning paradigm to study this mechanism. In this paradigm, participants are able to use information about long-term monetary reward to regulate their fear response to a pain-inducing stimulus.Many social and psychological interventions are based on giving people information to facilitate a change in the way they feel. However, the success of interventions that provide information is often varied. The aim of this research is to improve understanding of the process by which information changes emotional responses. Three studies employ repeated sequences of learning, intervention, and test stages, to examine the influence of verbal information on an emotional response. Study 1 will establish optimal parameters for the following two studies. Studies 2 and 3 will use functional magnetic resonance imaging to examine neural correlates in the brain between working memory, which maintains the information that is given as part of an intervention, and emotional response systems. Our objectives are to test three hypotheses: (1) information can generate emotions, and emotions so generated will resemble ‘natural’ emotions that are acquired through experience; (2) information can enhance and attenuate emotions; (3) emotional responses result from predictions that the brain makes about the impact of an event on a person’s goals. An improved understanding of the processes involved in successful interventions and behaviour change will be critical to a number of social objectives, particularly those related to health and well being.

Participants learned to fear inherently neutral face stimuli by taking part in a classical aversive conditioning paradigm with the factor threat anticipation (CS+/CS-). Two faces predicted physical pain, and two predicted safety (two CS+ faces: CS+1, CS+2, two CS- faces: CS-1, CS-2). In the counter-conditioning part of the experiment participants were instructed that while everything else remained the same and specifically the pain contingencies did not change, two of the face stimuli will now be followed by reward 50% of the time. Specifically, one of the faces that predicted pain (CS+1) and one that predicted safety (CS-1) also predicted that monetary reward will be added to participants’ account, and paid at the end of the experiment. The other two predicted that no reward will be delivered. The value of the reward was titrated individually so that it was equivalent to the (aversive) value of the painful stimulus, using the Becker–DeGroot–Marschak method. Because the amount of reward was chosen to be higher than the amount participants stated the pain was worth, CS+1 now predicted a net neutral overall utility. We used a factorial design to examine the interaction of pain and reward on expectancy while controlling for the main effects of threat and reward expectancies. Data: Participants performed the first part of the experiment outside the fMRI scanner, and the counterconditioning part of the experiment while they were scanned with fMRI. They provided ratings of their pain perception; how valuable the monetary reward was to them; to what degree they expected to receive pain or reward with each CS stimulus; how much they liked, or were threatened, by each CS stimulus. We collected their reaction time and accuracy on an incidental CS classification task, their skin conductance responses, and their structural MRI. Sample: All participants were healthy adult volunteers, screened for neurological and psychiatric conditions, who were not taking centrally-acting medications and had normal or corrected-to normal vision. Twenty five participants (13 male, mean age 23.1, S.D = 3.9) took part in the experiment. Data for one participant was excluded due to the participant’s failure to correctly demonstrate a consistent understanding of the CS-US contingencies during scanning . MR Data for a further 3 participants was excluded because of equipment failure. A further dataset was excluded because of motion artefact. This left 20 participants (11 male, mean age 23.3, S.D = 4.1) for which MR data was analysed.