Appetitive vs. Aversive conditioning in humans

Abstract

In classical conditioning, an initially neutral stimulus (conditioned stimulus, CS) becomes associated with a biologically salient event (unconditioned stimulus, US), which might be pain (aversive conditioning) or food (appetitive conditioning). After a few associations, the CS is able to initiate either defensive or consummatory responses, respectively. Contrary to aversive conditioning, appetitive conditioning is rarely investigated in humans, although its importance for normal and pathological behaviors (e.g., obesity, addiction) is undeniable. The present study intents to translate animal findings on appetitive conditioning to humans using food as an US. Thirty-three participants were investigated between 8 and 10 am without breakfast in order to assure that they felt hungry. During two acquisition phases, one geometrical shape (avCS+) predicted an aversive US (painful electric shock), another shape (appCS+) predicted an appetitive US (chocolate or salty pretzel according to the participants' preference), and a third shape (CS-) predicted neither US. In a extinction phase, these three shapes plus a novel shape (NEW) were presented again without US delivery. Valence and arousal ratings as well as startle and skin conductance (SCR) responses were collected as learning indices. We found successful aversive and appetitive conditioning. On the one hand, the avCS+ was rated as more negative and more arousing than the CS- and induced startle potentiation and enhanced SCR. On the other hand, the appCS+ was rated more positive than the CS- and induced startle attenuation and larger SCR. In summary, we successfully confirmed animal findings in (hungry) humans by demonstrating appetitive learning and normal aversive learning.

Keywords: classical conditioning; punishment; reward; skin conductance response; startle reflex.

Figure 1

Trials during the two acquisition phases (A) and the extinction phase (B). Participants learned that one shape (avCS+) predicted a mild painful electric shock, one shape (appCS+) predicted either a piece of chocolate or a small salty pretzel (according to their preference), and a third shape (CS–) did not predict any biologically salient event. Each shape was presented in conjunction with a picture depicting an electric shock, smarties/salty pretzel, or nothing depending on the association with the US. During the extinction phase, the three geometrical shapes were presented once again, but no USs were delivered. In addition, a fourth geometrical shape (NEW) was presented as a neutral control.

Figure 2

Ratings for valence (A) and arousal (B). Lines (with standard errors) depict the ratings after the habituation phase (T1), Acquisition 1 (T2), Acquisition 2 (T3), and the extinction phase (T4). The aversive CS+ (black solid line) acquired negative valence and high arousal after the two acquisition phases compared to the CS– (black dashed line) and the NEW (black dotted line). Importantly, the appetitive CS+ (gray solid line) acquired positive valence compared to the CS– and the NEW. ^*p < 0.05, ^**p > 0.01, ^***p < 0.001.

Figure 3

Startle responses (A) and skin conductance (B) (with standard errors) during the first acquisition phase (Acq1), the second acquisition phase (Acq2), and the extinction phase (Ext). Startle responses were significantly potentiated to the aversive CS+ (black solid line) and significantly attenuated to the appetitive CS+ (gray solid line) as compared to the CS– (black dashed line) during the acquisition phases. SCR was significantly greater to the avCS+ and the appCS+ compared to the CS–. No differences were revealed during the extinction phase. ^*p < 0.05, ^**p > 0.01, ^***p < 0.001.