Learning Through Consequences

  Behavioral conceptualizations and interventions are also based on the principle that organisms learn through the consequences of their actions. Principles of operant learning suggest that some consequences (i.e., those that are desired) will increase the frequency of a given behavior, or reinforce the behavior, and others (i.e., those that are not desired) will decrease the frequency of a given behavior, or punish the behavior. As such, any frequent behavior is assumed to be reliably associated with a consequence that is desired by the organism, even if there are also apparent consequences that might be assumed to be punishing. For example, a child who finds attention from a teacher to be desirable may continually act out in class because the attention she or he receives after acting out increases the frequency of the behavior, even though the teacher intends this attention to reduce the frequency of the behavior.

Several aspects of operant learning have important implications for understanding difficulties commonly seen in clients. First, immediate consequences generally influence behavior more strongly than do delayed consequences. Again, this makes sense from an evolutionary perspective because a consequence that is more proximal to a given action is more likely to be causally related to the action. However, this often leads to habitual behaviors that have destructive long-term consequences, despite immediately desirable consequences. Substance use is a classic illustration of these principles. Using substances typically has a number of initial consequences that are desirable, such as increased positive affect, decreased distress, and social connection. These are powerful contingencies that can make it very difficult to reduce the frequency of use, despite an individual’s awareness of a number of detrimental longer term effects of use (e.g., relational, occupational, and health-related problems). Associative learning also plays a role in substance use in that a wide range of cues come to be strongly associated with use and therefore elicit cravings that are difficult to ignore.

Reinforcers and punishers are defined by their effects on behavior (i.e., whether they increase or decrease the frequency of behavior, respectively). One cannot assume that a given consequence will necessarily act as either a reinforcer or a punisher because the value of a specific consequence is individually determined (and is likely based on learning history). Thus, certain consequences may be reinforcing for a given individual because they have previously been associated with other valued stimuli, but they may be meaningless, or punishing, for another individual. In the example presented earlier, for a child who commonly receives positive attention from adults, the negative attention of being scolded for misbehaving is less likely to be reinforcing and may have the punishing effect intended by the teacher. Certain facial expressions or tones of voice might be reinforcing for one individual who associates them with a loving parent but punishing for another individual whose abusive parent demonstrated similar modes of expression.

One aspect of operant terminology that is often confusing and is commonly misused is the terms positive and negative. Positive reinforcement refers to the presentation of something after a behavior that increases the likelihood that that behavior will recur (e.g., receiving a commission after selling a car is likely to increase sales over time), whereas negative reinforcement refers to the removal of something after a behavior that similarly increases the likelihood that that behavior will recur (e.g., a reduction in headache pain after taking medication is likely to increase the use of medication over time in response to headaches). Often, people erroneously use negative reinforcement when they mean punishment, or a consequence that makes it less likely a given behavior will recur.

Negative reinforcement is frequently a function of clinically relevant behaviors. Individuals often engage in behaviors that reduce or remove their distressing feelings or thoughts, thus negatively reinforcing the behavior and making it more likely they will engage in the behavior again (for an extensive review of experiential avoidance as a common function of clinical problems, see S. C. Hayes, Wilson, Gifford, Follette, & Strosahl, 1996). For instance, a natural behavioral response to threat cues is to escape or avoid them. This behavior is negatively reinforced by the immediate reduction in anxiety that an organism experiences, making it more likely that avoidance will recur. However, this avoidance also maintains fear by interfering with the extinction or inhibitory learning process described earlier. In this way, individuals can continue to experience anxiety in response to cues that are not truly indicators of threat, but have been associated with threat in the past, because their habitual avoidance impedes any new learning of a nonfearful association with the cue. As noted earlier, substance abuse may also be negatively reinforced by initially reducing or removing anxious, depressive, angry, or other distressing emotions, leading individuals to be more likely to drink or use drugs again. In fact, lapses in alcohol use after treatment are significantly associated with increases in negative affect (Witkiewitz & Aracelliz Villarroel, 2009). Similar functions have been proposed for eating-disordered behavior and deliberate self-harm (for a review, see Roemer & Orsillo, 2009).

Findings from experimental research have indicated that broader, flexible behavioral repertoires are more adaptive than narrowed, rigid repertoires (Drossel, Rummel, & Fisher, 2009). Broad, flexible repertoires allow for new learning to occur, so that an organism can function optimally even as contexts change so that consequences for a given behavior are different than they were during previous learning experiences. Clinical problems can be conceptualized as rigid, narrowed behavioral repertoires, and expansion of these repertoires is needed to help the individual engage in behaviors that will be naturally reinforced by his or her environment. For example, Joe learned to withdraw in response to communications of anger, which was adaptive growing up when his mother’s anger would escalate to violence. However, with his partner, he now withdraws when he detects even the slightest amount of annoyance, which his partner experiences as rejecting, escalating his negative affective experience and contributing to the cycle of the couple’s distress. If Joe can learn to expand his behavioral repertoire so that he can instead communicate directly when he perceives that his partner is annoyed, this new behavior may be reinforced by increased intimacy and cooperation, which would in turn decrease the couple’s distress. Thus, the new behavior would be both positively and negatively reinforced, so that it would gradually become more likely over time.

Cues are also relevant to operant learning. A discriminant stimulus that has been associated with a given behavior leading to a specific consequence becomes a cue to exhibit or inhibit the target behavior. In this way, individuals learn to behave in certain ways with a parent or teacher who is likely to reinforce certain behaviors and to refrain from other behaviors with a parent or teacher who is likely to punish them. Patterns of behaving that are insensitive to cues and contexts are particularly likely to be problematic and are often the source of clinical consultation. In the case of Joe, the behaviors he learned with his mother became a generalized pattern, and he failed to respond specifically to the cues present with his partner that indicated that a different behavior would be adaptive.

In working with Joe, the goal of treatment was to help him develop new behaviors that were sensitive to the novel discriminant stimulus of his partner. With other clients, the goal is to reduce the connection between the stimulus and the behavior. For example, many behaviors with an experiential avoidant function are cued by internal stimuli such as arousal, anxious thoughts, or negative affect. Over time, the individual learns to consistently respond to these cues with these behaviors, which are often followed immediately by at least slight decreases in distress, which further reinforce the behaviors. However, these behaviors can cause more long-term difficulties, including maintaining conditioned anxiety, as described earlier. Thus, treatment may focus on exposure to the cues (both internal and external) and prevention of the learned response, or response prevention. For instance, Jamila reported recurrent obsessive thoughts about germs causing disease in her family and habitually responded to these thoughts by cleaning compulsively. In therapy, she gradually learned to notice these thoughts and her urge to clean, but to refrain from cleaning in response to them. Over time, although she still had the thoughts from time to time, they were not as closely tied to the urge to clean, and she was able to engage in other aspects of her life more fully.

Early cue detection is an important element of new learning. As habitual patterns of responding are identified, the therapist and client use monitoring and review of the previous week to identify early cues for these behavioral patterns. When patterns of responding are identified earlier and earlier in the cycle, it becomes easier to practice alternative behaviors and responses and to develop new patterns of responding. For example, if a client notices the moment that he experiences tension in his shoulders, it is much easier to implement a relaxation strategy in that moment than if he waits until his anxious thoughts and reactions have begun to spiral into a much higher level of anxiety. New responses can be practiced at any point in the cycle, but earlier application is more likely to lead to desirable consequences, thus reinforcing the new behavior.

Behavior therapists typically emphasize reinforcement strategies as opposed to punishment strategies for a number of reasons (Drossel, Rummel, & Fisher, 2009). First, punishment is commonly associated with negative emotional responses, which can come to be associated with the person providing the punishment. As such, punishment may have a negative effect on the therapeutic relationship. Also, punishment only reduces the frequency of a behavior; it does not promote an alternative, adaptive behavior. Thus, punishment further narrows behavioral repertoires rather than expanding them. Each behavior occurs in a given context because its association to reinforcers is stronger than its association to punishers and because the association of alternative behaviors to reinforcers is less strong. Strengthening the reinforcing associations of an alternative, incompatible behavior will therefore also decrease the likelihood of the target behavior without introducing the less desirable correlates of punishment.

As noted earlier, the nature of the reinforcers that will be influential is specific to individuals and may also be specific to given contexts. A careful functional analysis is necessary to determine the consequences that are likely to increase the frequency of a desired behavior. Natural reinforcers will be more powerful than artificial or contrived reinforcers because they are more likely to occur in the client’s environment and will therefore help maintain the new behavior outside of therapy. Social reinforcers—or attention, acknowledgment, praise, and approval from other people—are particularly powerful forms of reinforcement because people naturally seek these consequences and they occur naturally in people’s environments (Spiegler & Guevremont, 2010). However, Kohlenberg and Tsai (1995) provided a useful clinical example of how social reinforcers in therapy can also be contrived or artificial. When working with a client who has difficulty expressing his anger, a therapist might be inclined to respond to an expression of anger by smiling and saying, “I’m so glad you shared those feelings with me.” However, it is very unlikely that anyone in the client’s life will respond similarly to this kind of communication. However, if the therapist attends to the source of the client’s anger and makes an attempt to address it accordingly, this more natural contingency might also be maintained in the client’s life.

Therapy can be used to increase the reinforcing value of behaviors when the positive consequences of these behaviors are more long term and therefore do not automatically influence behaviors. From a behavioral perspective, this alteration of the reinforcement value of behaviors is a way of enhancing motivation (Drossel, Rummel, & Fisher, 2009). For instance, motivational interviewing (Miller & Rollnick, 2002) helps the client to identify the ways in which current behaviors are inconsistent with longer term goals or values. This process will likely serve to decrease the reinforcing properties of goal-inconsistent behaviors (e.g., avoiding social interactions or drinking excessively) while increasing the reinforcing properties of goal-consistent behaviors (e.g., initiating social contact or choosing not to take a drink). Similarly, values clarification, an element of acceptance and commitment therapy and other acceptance-based behavior therapies (e.g., K. G. Wilson & Murrell, 2004), helps clients to identify what is personally meaningful to them (e.g., being responsive to a family member or pursuing work that is intellectually challenging). This connection to meaning can enhance the reinforcing properties of behaviors that are otherwise not as strongly immediately reinforcing (e.g., sitting down to write) while refraining from behaviors that are more readily reinforced but less valued by the individual.

Interestingly, people’s own behavior can serve as a reinforcer for other behaviors. The Premack principle refers to the consistent finding that high-frequency behaviors that occur reliably after the occurrence of low-frequency behaviors will increase the frequency of those low-frequency behaviors (Premack, 1965). For example, if while writing this book we found that sitting down to write had become a low-frequency behavior and checking e-mail was a high-frequency behavior, if we intentionally only check e-mail after having written for 60 minutes, checking e-mail would reinforce our writing behavior and we would begin to sit down to write more frequently. Spiegler and Guevremont (2010) presented a case example of a psychiatric patient who was not engaging in any social behaviors and whose only frequent behavior was sitting in a chair. When the hospital staff introduced contingencies so that she could sit in the chair only after engaging in some sort of social interaction and gradually increased the length of social interaction required for the chair sitting to occur (beginning with 2 minutes), within 3 weeks the patient was spending more time socializing than sitting in the chair.

Different schedules of reinforcement yield different patterns of responding. Continuous reinforcement, or reinforcing a behavior each time it occurs, is most helpful when a new behavior is being learned. However, once a behavior is established, an intermittent reinforcement schedule, or only reinforcing a behavior some of the times it occurs, leads to more robust behavior that is more likely to continue over time and in multiple contexts. An intermittent schedule is similar to the reinforcement that occurs naturally in people’s lives. Often, the persistence of a problematic behavior can be understood by past experiences of intermittent reinforcement. For instance, people often have a difficult time ending relationships that are occasionally extremely positive (and thus powerfully reinforcing) despite extended times when they are negative (and thus not reinforcing at all). This phenomenon may help explain people remaining in abusive relationships—the “honeymoon period” after an abusive outburst is often powerfully reinforcing, leading the victim to stay in the relationship.

An individual’s ability to exhibit a given behavior is also an important consideration in trying to expand a client’s behavioral repertoire. Often, the learning histories that have narrowed behavioral repertoires have also failed to promote learning of important skills, such as emotional awareness and regulation (cf. Linehan, 1993a) or social skills. These skills deficits may further impede new learning because it is difficult for the client to act in a way that is naturally reinforced by her or his environment. In these cases, instruction in skills can be an important part of therapy (as described in Chapter 4). However, humans naturally respond to subtle discriminative stimuli in social situations that cannot be captured in instruction (Drossel, Garrison-Diehn, & Fisher, 2009), so behavioral learning remains an important aspect of skills training. In addition, natural reinforcers for performing the skills (e.g., expressions of understanding when a client describes an emotional experience) will be more powerful than artificial reinforcers (e.g., verbal praise), although artificial reinforcers may initially be useful for establishing the newly learned behavior.