Conditioned Frustration: Behavior and Brain Mechanisms

Frustration is induced when a signal previously paired with a more preferred reward is surprisingly paired with a less preferred one. Frustration is not aroused when a poor outcome is anticipated, but when reward downshift occurs unexpectedly. For example, a student who has prepared for an exam would feel more frustrated by a poor grade than a student who did not prepare for the exam. Like fear, frustration is an aversive internal response with behavioral and physiological consequences (e.g., increased aggressive behavior and stress hormones, among others; Papini & Dudley, 1997). When the effects are measured within seconds to minutes of the frustrating event, the emotion is referred to as primary frustration—an unconditioned response to a surprising reward downshift. Frustration can also be a conditioned response—conditioned frustration. After studying hard and failing in one exam, a student can feel uneasy before a second exam, even after studying the required material. The student is anticipating the possibility of a second failure. One consequence of a downshift in reward amount or quality is that instrumental behavior deteriorates beyond the level of a control group exposed only to the less preferred reward. This is called successive negative contrast (SNC) and the basic design is shown in Table 9.4 (see Figures 7.15 and 7.16). The SNC effect has been studied with instrumental (iSNC; e.g., running in a runway) and consum-matory (cSNC; intake of sucrose solutions) procedures.

An experiment with 3-month-old babies illustrates iSNC with an ingenious procedure (Mast, Fagen, Rovee-Collier, & Sullivan, 1980). A mobile with colorful wooden blocks was suspended over the crib, and a ribbon connected this mobile to the baby’s right ankle. By kicking their right legs (response), babies could control the amount of movement of this

Table 9.4 Design of a successive negative contrast experiment

Note: Groups differ during preshift sessions, but receive the same treatment during postshift sessions. The response can be instrumental (kicking, running) or consummatory (licking, chewing). The rewards can vary in terms of magnitude (large vs. small) or quality (preferred vs. less preferred). Importantly, the smaller or less preferred reward is acceptable in the absence of experience with the larger or more preferred reward.

Early Learning and Behavior 343 mobile (reward). Babies kicked more frequently when the mobile had 10 colorful wooden blocks, than when it had only 2 blocks. Interestingly, an unexpected downshift from 10 to 2 blocks caused an increase in the frequency of negative vocalizations, including crying and fussing, indicating the aversive nature of the change. Moreover, after the 1(02 reward downshift, kicking frequency increased during the initial 9 min and then decreased below the level of the 2-^2 unshifted control group.

Infant rats show a graded emergence of both the iSNC and cSNC effects. Rats ran (response) down a runway to obtain different amounts of milk (reward) in the goal box. Infants received two sessions of training at postnatal days 16-17, 20-21, or 25-26. Downshifted infants received 0.3 ml of milk (large reward) in preshift trials, but 0.02 ml of milk (small reward) in postshift trials. Unshifted controls were always rewarded with 0.02 ml of milk (small reward). The speed of running increased over trials at all ages, thus showing equivalent learning skills. Moreover, infants in all ages ran faster for the large reward than for the small reward, providing evidence that the magnitudes were discriminable (Figure 9.10). However, reward downshift induced no behavioral change in 17-day-olds, a hint of iSNC in 21-day-olds, and a significant iSNC effect in 26-day-old rats (Chen, Gross, & Amsel, 1981). The iSNC effect is evident in the undershooting of running speed during the postshift phase. Similarly, the cSNC effect is evident in 18-day-old rats, but not in 14- or 17-day-old infants (Suarez, Mustaca, Pautassi, & Kamenetzky, 2014).

The SNC and other situations involving surprising nonreward emerge gradually in infant rats between postnatal days 12 and 26 (Amsel, 1992; Table 9.5). The instrumental behavior of infant rats before the emergence of these learning phenomena resembles that of adult rats with lesions in some brain areas, such as the hippocampus. Interestingly, during the same period when these appetitive learning effects emerge, the dentate gyrus of the hippocampus is undergoing rapid differentiation of its population of granular cells. When differentiation was disrupted by exposure to X-rays between postnatal days 2 and 15, then the emergence of some of these learning phenomena was delayed. For example, the PREE was eliminated in 21-day-old rats. Thus, rats receiving 50% partial reinforcement extinguished the running

Figure 9.10 Emergence of instrumental successive negative contrast (iSNC) in infant rats. See Table 9.4 for a description of the design. Two groups received training across two phases. The critical feature is the decrease in the amount of milk reward in the downshifted group from Phase 1 to Phase 2. Whereas both groups are treated identically in Phase 2, the running performance of 25- to 26-day-olds deteriorates significantly below that of the unshifted controls, thus showing the iSNC effect. Younger rats show no evidence of an iSNC effect. However, rats of all ages learn faster with the large reward than with the small reward, suggesting that the failure of iSNC to emerge in younger rats is not caused by an inability to perceive the difference in reward magnitude.

Source: Chen, Gross, and Amsel (1981); reproduced with permission from the American Psychological Association.

Table 9.5 Age of appearance of learning phenomena involving surprising nonreward in infant rats

Note: This developmental information applies to the instrumental version of the SNC effect (e.g., running responses in a runway).

response at the same rate than rats trained under continuous reinforcement, rather than more slowly as did control infants without lesions in the dentate gyrus (Diaz-Granados, Greene, & Amsel, 1992). The dependence of the iSNC on hippocampal function resembles the development of contextual fear conditioning discussed earlier.