Drawn in by the unpredictable: reward uncertainty broadens the spotlight of incentive salience attribution

The motivational value assigned to cues (CSs) associated with reward (UCS) is known as incentive salience, and is determined through the synergy of previous learning of the CS-UCS association (the sum of experiences an individual has with that cue-reward combination) and the mesolimbic state of the individual at the time of cue re-encounter. Typically, the amount of incentive salience attributed to a cue is believed to increase with the relative reliability that the cue predicts reward, where cues that consistently predict reward become powerful motivational magnets, eliciting approach and even consummatory behaviors (similar to those produced by the reward itself). In contrast, cues that predict reward with a lesser degree of certainty should theoretically be less attractive, and accrue lower levels of incentive salience. However, in some cases, cues that predict reward with maximal levels of uncertainty, such as in gambling, may generate higher levels of incentive salience attribution and become very attractive. In order to examine the impact of uncertainty, we used a Pavlovian conditioned approach task (autoshaping), where a bright noisy lever cue precedes delivery of a sucrose reward pellet. We have previously shown that a cue that predicts a reward only half the time (probability = 50%), but where the value of that reward can vary (1, 2 or 3 sucrose pellets), attracts more incentive value than a cue that systematically (probability = 100%) predicts a single sucrose reward pellet. Here we examined whether the increased incentive value attributed to uncertain cues was equally distributed across cues that were proximal (5 cm) or distal (20 cm) to the site of reward (magazine). Our results confirm that incentive value is greater for uncertain cues, and show that it is equally distributed across proximal and distal cues. In contrast, when the CS cue predicts the UCS reward with absolute certainty, incentive value is preferentially given to proximal cues, with distal cues being almost ignored. This suggests that reward uncertainty not only attributes more incentive value to cues, but that it may also act by recruiting a broader range of cues. However, since the learnt predictive value of the cue is low, this increase in incentive salience may be generated through greater mesolimbic activity driven by anticipation and/or stress of an unpredictable reward. In conclusion, reward uncertainty may transform a broad array of cues into powerful motivational magnets, which in turn could help explain what makes gambling so attractive.