Michael & Miguel ARTICLE 2020

The scenario above suggests that the value of certain stimuli may vary conditionally with other environmental factors. Mexican food is not wanted all the time, and behaviors that produce it (e.g., driving to the restaurant) vary depending on someone’s motivation. From a behavior analytic perspective, if a certain behavior is not occurring, there is a chance that reinforcing consequences are not being produced. However, even if consequences are available, their value or effectiveness can vary over time. Thus, the concept of reinforcement alone cannot properly explain a person’s motivation to act (Michael, 1993). Going to a restaurant and ordering food invariably produces food. However, the likelihood of going to a restaurant may vary depending on one’s motivation to eat. When we say someone is motivated or wants to do something, we are assuming that (a) what is wanted would function as a reinforcer at that moment, and (b) the current frequency of any behavior that has previously been so reinforced will increase. This chapter discusses motivation from a behavior analytic perspective.

DEFINITION AND CHARACTERISTICS OF MOTIVATING OPERATIONS Early in his career, Skinner (1938) used the term drive to describe changes in behavior that were a function of motivational variables.1 Although the term appeared in Keller and Schoenfeld’s (1950) didactic treatment of Skinner’s work, these authors placed an emphasis on the specific environmental operations (i.e., deprivation, satiation, and aversive stimulation) that are necessary to establish the drive (Miguel, 2013). Although not conforming to the same usage set forth by Keller and Schoenfeld, the term establishing operation (EO) was reintroduced by Michael (1982, 1993) to describe any environmental variable that momentarily alters (a) the effectiveness of some stimulus, object, or event as a reinforcer; and (b) the current frequency of all behavior that has been reinforced by that stimulus, object, or event. Subsequently, Laraway, Snycerski, Michael, and Poling (2003) proposed the inclusive term motivating operation (MO) to encompass the fact that these variables serve not only to strengthen, but also to weaken, the value of a consequence as a reinforcer.2 The term setting event has also been used to describe some of the variables that may fit under the MO definition (see Box 16.1: “What’s a Setting Event?”). Thus, an MO can be defined as an environmental variable that has two effects: value-altering and behavior-altering effects. The value-altering effect is either (a) an increase in the reinforcing effectiveness of some stimulus, object, or event, in which case the MO is an establishing operation (EO); or (b) a decrease in reinforcing effectiveness, in which case the MO is an abolishing operation (AO). The behavior-altering effect is either (a) an increase in the current frequency of behavior that has been reinforced by some stimulus, object, or event, called an evocative effect; or (b) a decrease in the current frequency of behavior that has been reinforced by some stimulus, object, or event, called an abative effect. In addition to frequency, other aspects of behavior such as response magnitude, latency, and relative frequency (occurrence per opportunity) can be altered by an MO. These relations are shown in Figure 16.1. Food deprivation is an EO that increases the effectiveness of food as a reinforcer (value-altering effect) and evokes all behavior that has been reinforced by food (behavior-altering effect). In other words, after not having food for a while, food becomes a reinforcer, and behaviors such as looking for food are increased. Food ingestion (consuming food) is an AO that decreases the effectiveness of food as a reinforcer (value-altering effect) and abates all behavior that has been followed by food reinforcement (behavior-altering effect). In other words, after eating, the reinforcing effectiveness of food is decreased, and so is looking for food (along with other previously learned behaviors that have produced food). These changes of behavior are said to be momentary. For instance, food deprivation (MO) increases the value of food, as well as all behaviors that have led to food, for as long as the organism is deprived. As food is obtained, the effectiveness of food deprivation as an MO is decreased (the animal becomes less hungry), as evidenced by a reduction in (a) the value of food as a reinforcer and (b) food-seeking behaviors. Although the term hunger has been used to describe the effects of food deprivation, the term is simply a description (i.e., tact) of physiological changes that may accompany food deprivation (e.g., hunger pangs). Although these physiological sensations may prompt someone to get food, deprivation operations have direct value-altering and behavior-altering effects (Miguel, 2013). Thus, terms such as hunger and thirst are unnecessary (intervening) variables that may obscure a behavior analysis of motivation. The effects of the MO related to food are shown in Figure 16.2. Painful stimulation is an EO that increases the effectiveness of pain reduction as a reinforcer and evokes all behavior that has been reinforced by pain reduction. So, a headache would make its reduction (or elimination) an effective reinforcer, and would also evoke behaviors, such as taking an aspirin, that have produced headache removal in the past. Of course, these two effects would only be observed while the organism is experiencing a headache, and for this reason we refer to the effects of the MO as being momentary. Conversely, a decrease in painful stimulation is an AO that decreases the effectiveness of pain reduction as a reinforcer and abates all behavior that has been followed by pain reduction. In other words, once the headache is gone, its elimination no longer functions as a reinforcer, and the likelihood of taking an aspirin is greatly reduced. The effects of the MO related to painful stimulation are shown in Figure 16.3. Most of the statements in this chapter about value-altering and behavior-altering effects refer to relations involving reinforcement rather than punishment. It is reasonable to assume that MOs also alter the effectiveness of stimuli, objects, and events as punishers, with either establishing or abolishing effects; and also alter the current frequency of all behavior that has been punished by those stimuli, objects, or events, either abating or evoking such behavior. Later in the chapter we will consider MOs for punishment. Isolating Value- and Behavior-altering Effects The value-altering effect of an MO influences the potency of a reinforcer to increase the future frequency of behavior (Laraway, Snycerski, Olson, Becker, & Poling, 2014). An MO’s value-altering effect is observed when the manipulation of certain environmental variables (e.g., 4, 8, and 12 hours of food deprivation) affects the speed with which an organism acquires new behaviors that produce the relevant consequence (e.g., food). In this way, an MO directly changes the potency of reinforcers (or punishers) to increase and maintain (or decrease and eliminate) behavior (Laraway et al., 2014). The value of a reinforcer may also vary depending upon specific response requirements (DeLeon, Frank, Gregory, & Allman, 2009). One way to assess the potency of a specific reinforcer is to use a progressive ratio (PR) schedule in which response requirements increase after the delivery of each reinforcer (Roane, 2008; see Chapter 13). Under this arrangement, an organism may respond under high schedule requirements to obtain one consequence, but not another, even when both consequences have been observed to increase behavior under low response requirements (Roane, Lerman, & Vorndran, 2001). This procedure may serve to measure the momentary value of a reinforcer whose effectiveness varies depending on the MO. Other potential measures of reinforcer potency may include choice preparations, and latency to respond during a discrete trial (Poling, Lotfizadeh, & Edwards, 2017). The behavior-altering effect is interpreted as a change in frequency due to the organism’s behavior contacting reinforcement, implying that this change occurs only after the reinforcer has been obtained. However, a strong relation exists between the level of the MO and responding under extinction—when no reinforcers are being received (Keller & Schoenfeld, 1950). Early research (Skinner, 1938) showed that rats that had learned to press a lever to obtain food, when exposed to an extinction condition at different levels of deprivation, pressed the lever at different rates according to their levels of deprivation. Thus, different deprivation levels differentially evoked responses that had been previously learned. Similarly, if a child’s tantrum has been followed by a parent’s attention, the duration and intensity of the tantrum may be correlated with the amount of attention that the child has received before the tantrum. The value-altering and behavior-altering effects of an MO occur simultaneously but are considered independent in the sense that one does not derive from the other. Food deprivation, for instance, increases behaviors that have been previously reinforced by food prior to obtaining any food (evocative effect). Food deprivation would also increase the reinforcing potency of food (the establishing effect). Thus, food-seeking behaviors would increase further after contacting food (Michael, 2000). Unfortunately, once behavior contacts reinforcement, these two effects may become difficult to discern (Poling et al., 2017). Direct and Indirect Effects of Motivating Operations The behavior-altering effect of an MO can increase (or decrease) the frequency of behavior because of a direct evocative (or abative) effect. The strength of the MO seems to be correlated with its capacity to directly evoke previously learned behavior. For example, when food deprivation is high, food-seeking behaviors may occur in situations that differ from the situations in which those behaviors were reinforced in the past (Lotfizadeh, Edwards, Redner, & Poling, 2012). An individual who has not eaten for a few hours is likely to request a snack from an acquaintance who usually carries some in her backpack (SD). However, under extreme food deprivation, the individual might ask strangers for a snack, even though there has been no previous history of obtaining food from them. It seems that “the range of stimuli that evokes responding is wider when deprivation is high than when it is lower” (Lotfizadeh et al., 2012, p. 98). In other words, strong MOs can directly evoke behavior in the absence of relevant discriminative stimuli or, at least, in the presence of untrained stimuli. Thus, the direct evocative effect of the MO may be described as its capacity to evoke behavior irrespective of relevant discriminative stimuli.3 It is important to remember that individuals do not need to be “aware” of the effects of reinforcement for it to affect behavior. Likewise, the effectiveness of MOs does not depend on individuals being able to describe the effects of MOs upon their behavior (see Box 16.2, “Do You Need to Know You Are Motivated?”). The MO indirectly affects the evocative or abative strength of relevant discriminative stimuli (SDs). Food deprivation, for instance, would increase the discriminative control of any stimulus that has been correlated with the availability of food. Therefore, the friend who carries snacks in her backpack would serve as a stronger SD that would evoke a request when food deprivation is high. Also, the stronger the headache (MO), the more likely it is that the sight of the aspirin (as an SD) would evoke the behavior of taking it. Note that the value of reinforcers present during discrimination training also depends upon a current MO (Laraway et al., 2003). In other words, the establishment of stimulus control over behavior not only depends on but also varies as a function of MOs. For instance, a child would never learn to discriminate between blue and red if no effective reinforcer were available for correct responses. To summarize, in the absence of an MO that establishes the value of the consequence as a reinforcer, discrimination learning is unlikely to occur, and a previously effective SD would also not evoke behavior. The value-altering effect also indirectly influences the reinforcing effectiveness of relevant conditioned stimuli. For example, food deprivation not only increases the capacity of food as a consequence to strengthen behavior (direct establishing effect), but also enhances the reinforcing effectiveness of all other stimuli that have been correlated with food, such as the smell of food, the sight of a restaurant, and so on. In the same way, a headache would not only establish its removal as an effective reinforcer, but also all stimuli that have been paired with pain reduction, such as an aspirin, a nap, a dark environment, and so forth. The latter effect, a conditioned one, will be discussed in the section on conditioned MOs, below. Figure 16.4 displays the different direct and indirect effects of the MO. Behavior-altering versus Function-altering Effects4 Both MOs and SDs are antecedent variables that have behavior-altering effects. Antecedents can evoke or abate responses, but their occurrence does not alter the organism’s functional behavioral repertoire. Thus, as antecedent variables, MOs alter the current frequency of all behaviors relevant to that MO. For example, a difficult task may serve as an establishing operation that momentarily increases all behaviors that in the past have produced removal of the task. Presenting a difficult task to a child who has learned that tantrums often result in task postponement or removal is likely to produce a tantrum (i.e., evocative effect). Conversely, if the task is not presented, or made easier (an abolishing operation), then the likelihood of tantrums is diminished (i.e., abative effect). However, as an intervention strategy, the removal of the task does not alter the child’s repertoire; it only momentarily abates behavior that would otherwise occur if the motivating variable (difficult task) were in place. In other words, the child will not behave differently in the presence of difficult tasks solely as a result of antecedent 4 Michael (1986, 2004) used the terms “repertoire-altering effect” and “function-altering effect” to classify operations (e.g., reinforcement) that would change the future probability of an organism’s behavior, as well as alter the function of stimuli. manipulations.5 Thus, reducing or eliminating the MO for a specific problem behavior is a temporary rather than permanent solution. When the MO is again in place, the problem behavior may re-occur (Michael, 2000). Nonetheless, eliminating environmental conditions (e.g., MOs) that contribute to the occurrence of problem behavior may be necessary,6 especially when these conditions are impoverished or aversive (McGill, 1999). Consequences can change the organism’s repertoire, enabling the organism to behave differently in the future. Thus, they have function-altering effects consequence variables include reinforcers, punishers, and the occurrence of a response without its reinforcer (extinction procedure) or without its punisher (recovery from punishment procedure). Consequences alter the future frequency of whatever behavior immediately preceded those consequences. In the previous example, if task removal or reduction of task difficulty is made contingent upon the child asking for a break or help, the frequency of these behaviors would increase under similar conditions in the future. In other words, the child’s functional repertoire is changed as he or she can engage in two new responses: asking for a break or asking for help.

DISTINGUISHING BETWEEN MOs AND SDs As described above, MOs and SDs are both antecedent variables that alter the current frequency of some particular behavior (i.e., evoke). Each operant variable influences response frequency because of its relation to reinforcing or punishing consequences. An SD affects behavior because its presence has been correlated with the differential availability of an effective reinforcer in the past. Differential availability means the relevant consequence has been available in the presence of the SD (i.e., reinforcement) and unavailable in its absence (i.e., extinction; see Chapter 24). For instance, if a food-deprived rat receives food for lever presses in the presence, but not absence, of a light, the light will eventually evoke lever presses. In this case, the light would be functioning as an SD, since food delivery was more available in its presence than in its absence. Note that although food was never delivered when the light was off (the S∆ condition), food would have served as an effective reinforcer had it been delivered when the light was off. In other words, food was valued (i.e., reinforcing) in both conditions, but available only when the light was on. The reduced frequency of lever presses during the S∆ condition is due to extinction. An MO, in contrast, controls behavior because of its relation to the differential effectiveness of a reinforcer for that behavior. Differential effectiveness means that the relevant consequence has been effective in the presence of, and ineffective in the absence of, the MO. In the example above, the rat would press the lever only in the presence of the light (SD) when the MO is present (i.e., food deprived) and not when the MO is absent (i.e., food satiated). When the MO is absent, behavior decreases not because the reinforcer is unavailable (extinction), but because it is no longer valuable or effective. Now imagine that the rat learns to press the lever to eliminate shock. In this case, pressing the lever produces pain reduction (or elimination) as a form of reinforcement. If painful stimulation as an antecedent were to be considered an SD, pain reduction (the reinforcer) must have been available as a consequence in the presence of painful stimulation, but not in its absence. More importantly, like food in the previous example, pain reduction would have been effective as a reinforcer in both conditions, regardless of its availability. It is with respect to this requirement that some motivating variables fail to qualify as discriminative stimuli. Pain reduction could not serve as a reinforcer in the absence of pain as an antecedent.7 In other words, in the absence of painful stimulation, behavior has not decreased due to extinction, but because there is no possible valuable reinforcer. The organism is not motivated to engage in behavior to reduce pain, if no pain is being felt. In the previous example of light-on/light-off discrimination training, the reinforcer is available in the light-on condition 7 It is true that pain reduction as a form of reinforcement is unavailable in the absence of pain, and also, food reinforcement is, in a sense, unavailable in the absence of food deprivation, but this is not the kind of unavailability that occurs in discrimination training and that develops the evocative effect of an SD (Michael, 1982). and unavailable, yet still effective, in the light-off condition. In other words, the rat still “wants food” when the light is off but food is not available. The behavior occurs less frequently in the light-off (S∆) condition because it no longer produces an effective reinforcer. In contrast, in the absence of food deprivation or painful stimulation, food or pain removal, respectively, is not an effective reinforcer. Hence, food deprivation and painful stimulation qualify as MOs because they momentarily alter (a) the effectiveness of some stimuli, objects, or events as reinforcers (value-altering effect); and (b) the frequency of the behavior that has been reinforced by those stimuli, objects, or events (behavioraltering effect). Similarly, an itch does not serve as a discriminative stimulus that evokes scratching it. Although the reinforcer (removing the itch) is available in the presence of the itch and unavailable in the absence of the itch, this is not an effective reinforcer in the absence of the itch. In this sense, the itch is best interpreted as functioning as an MO that establishes its removal as an effective reinforcer, and momentarily evokes scratching. In other words, having an itch is what motivates you to scratch. Many antecedent stimuli that have been considered to evoke behavior as SDs would be better classified as MOs. This is particularly the case with aversive stimuli. For example, a headache is an MO that directly establishes headache removal as an unconditioned reinforcer and indirectly establishes aspirin as a conditioned reinforcer (see Figure 16.5). A headache also evokes behaviors that have produced pain reduction (finding and taking aspirin) in the past. If aspirin is found in the bathroom but not in the kitchen, then the bathroom will function as an SD for the behavior of looking for aspirin because it has been correlated with availability of an effective reinforcer, while the kitchen will function as an S∆ for looking for aspirin because it has been correlated with the unavailability of an effective reinforcer. In other words, for discrimination to take place, an individual needs to find aspirin in the bathroom, and not find aspirin in the kitchen, respectively, when needed (MO; see first two rows of Figure 16.5). In the absence of the headache (AO), one can still find aspirin in the bathroom, but there is no reason to look for it (available, but not a reinforcer), since there is no headache to be removed (unavailable non-reinforcer). Thus, in the absence of the headache, looking for an aspirin will not occur in either location (see bottom row of Figure 16.5). In applied settings, when a child is given an instruction to complete a difficult task, the reinforcer may be task completion or termination. Additionally, the instruction would also acquire conditioned aversive properties due to its correlation with difficult tasks. Thus, the instruction could serve as an MO that establishes its removal as a reinforcer and evokes avoidance behavior (Carbone, Morgenstern, Zecchin-Tirri, & Kolberg, 2010). In contrast, if task completion has produced praise as a reinforcer in the presence, but not the absence, of the instruction, then the instruction may be functioning as an SD. Other events best classified as MOs rather than SDs will be discussed later in this chapter, most specifically in the section on conditioned motivating operations. Although behavior analysts use the three-term contingency (antecedent-behavior-consequence) to understand operant behavior, any given response may be evoked by a multitude of variables functioning as antecedents (i.e., multiple control). A child who enters the kitchen and asks for a cookie does so because in the past such behavior has produced reinforcement under similar circumstances. These circumstances involve variables that establish cookies as reinforcers, as well as increase the likelihood of asking for them (food deprivation as an MO). However, the behavior is more likely to occur in environments where cookies are available (kitchen as an SD) and in the presence of an adult (also an SD) who can give the child a cookie (Skinner, 1957). In this case, food deprivation is correlated with increasing the effectiveness of cookies as reinforcers, while the kitchen and the adult are both correlated with the availability of cookies. To summarize, SDs alter the differential availability of a currently effective form of reinforcement for a particular type of behavior; MOs alter the differential reinforcing effectiveness of a particular type of environmental event. In nontechnical terms, an SD tells you that something you want is available; an MO makes you want something. UNCONDITIONED MOTIVATING OPERATIONS (UMOs) For all organisms there are events, operations, and stimulus conditions with unconditioned value-altering effects. In other words, these MOs establish or abolish the value of stimuli as reinforcers in the absence of prior learning. For example, humans are born with the capacity to be affected by food reinforcement as a result of food deprivation and by pain reduction reinforcement as a result of pain onset or increase. Thus, food deprivation and painful stimulation are unconditioned motivating operations (UMOs). 8 MOs are classified as unconditioned based upon the unlearned aspect of their value-altering effects, since the behavior-altering effects of MOs are usually learned. Said another way, we are born with the capacity to be affected by food reinforcement as a result of food deprivation, but we learn the behaviors evoked by food deprivation—asking for food, going to where food is kept, cooking, and so forth. Deprivation of food, water, oxygen, activity, and sleep all have reinforcer-establishing and evocative effects (see Table 16.1). For instance, water deprivation momentarily establishes the effectiveness of water as a reinforcer and evokes all behaviors that have produced water. By contrast, food and water ingestion, oxygen intake, being active, and sleeping have reinforcer-abolishing and abative effects (see Table 16.2). For instance, water ingestion abolishes the effectiveness of water as a reinforcer and abates all behaviors that have produced water. The passage of time since last sexual activity functions as a UMO relevant to sexual stimulation. This form of deprivation establishes the effectiveness of sexual stimulation as a reinforcer and evokes behavior that has produced such stimulation. In contrast, sexual orgasm is a UMO that abolishes (weakens) the effectiveness of sexual stimulation as a reinforcer and abates (decreases the frequency of) behavior that has achieved that kind of reinforcement. In addition, tactile stimulation of erogenous regions of the body seems to function as a UMO in making further similar stimulation even more effective as a reinforcer, and in evoking all behavior that in the past has achieved such stimulation. UMOs related to temperature changes include becoming uncomfortably cold and uncomfortably warm. Getting cold is a UMO that establishes getting warmer as a reinforcer and evokes behavior that has produced warmth, such as putting on a jacket or moving closer to a heat source. A return to a comfortable (i.e., nonaversive) temperature is a UMO that abolishes increased warmth as a reinforcer and abates behavior that has produced warmth. In this case, behaviors such as putting on a jacket are less likely to occur. Getting uncomfortably warm is a UMO that establishes a cooler temperature as an effective reinforcer and evokes behaviors that have resulted in a body-cooling effect. For example, getting warm may evoke the behavior of turning on the air conditioning. A return to a normal temperature abolishes being cooler as a reinforcer and abates body-cooling behavior. In this case, turning on the air conditioning is less likely to occur. An increase in painful stimulation establishes pain reduction as a reinforcer and evokes the (escape) behavior that has achieved such reduction. As described previously, a headache establishes its own removal as a reinforcer and evokes any behavior that has eliminated headaches, such as taking aspirin. A decrease in painful stimulation abolishes the effectiveness of pain reduction as a reinforcer and abates the behavior that has been reinforced by pain reduction. Thus, when a headache is gone, headache removal is no longer a reinforcer and taking an aspirin is less likely to occur. In addition to establishing pain reduction as a reinforcer and evoking the behavior that has produced pain reduction, painful stimulation may evoke aggressive behavior toward another organism.9 Painful stimulation may function as a UMO that establishes events such as signs of damage as effective reinforcers and evokes behaviors that have been reinforced by such signs. Skinner (1953) made this case in his analysis of anger and extended it to the emotions of love and fear.10 Although it is indisputable that both food deprivation/satiation and painful stimulation/reduction have unlearned value-altering effects, restriction of and access to certain conditioned reinforcers also seem to affect their value. This is the case with social reinforcers, such as praise (Vollmer & Iwata, 1991). It could be argued that for infants, attention (e.g., touch, parent’s voices) is a form of unconditioned reinforcer whose value may be affected by deprivation and satiation. However, for most adults, attention functions as a conditioned reinforcer due to its relation to other forms of effective reinforcers (Michael, 2000). In this case, access and restriction to attention (or any other conditioned reinforcer) would function as a type of conditioned MO, as discussed below. MOs FOR PUNISHMENT MOs may also alter (increase or decrease) the punishing effectiveness of a stimulus, as well as the frequency (evoke or abate) of behaviors that have been punished by that stimulus. Thus, an EO related to punishment would have a punisher-establishing effect and an abative effect. For example, a migraine would establish the effectiveness of bright light as a punisher and decrease the frequency of all behaviors that in the past have produced the bright light (e.g., opening the curtains, turning a light on). 9 Some of this aggression may be elicited by pain functioning as a respondent unconditioned stimulus (Ulrich & Azrin, 1962). 10For a discussion of Skinner’s approach to emotional predispositions in the general context of MOs, see Michael (1993, p. 197). Some biological states, such as pregnancy, may establish certain tastes or smells as punishers and decrease the occurrence of behaviors that have produced them. An AO related to punishment would have a punisher-abolishing effect and an evocative effect, in that behaviors that have been punished would occur again (assuming that a reinforcer contingency for the previously punished behavior is intact). Ingesting alcohol may decrease the effectiveness of social disapproval as a punisher and increase behaviors that have produced some social embarrassment (assuming those behaviors also produced some reinforcement in the past). Most punishers that affect humans are effective because of a learning history; that is, they are conditioned rather than unconditioned punishers. In the case of conditioned punishers that were established by having been paired with the removal (or reduced availability) of reinforcers (i.e., negative punishment), their effectiveness as conditioned punishers will depend on the same MOs that establish the value of the reinforcers that were removed. For instance, a parent’s disapproving look may function as a conditioned punisher as a result of having been paired with restricted access to video games. The disapproving look will be most effective as a conditioned punisher when an MO for playing video games is in effect. If the child has already played enough video games for the day (i.e., is “satiated”), then the disapproving look will be less effective in reducing behavior. Thus, social disapproval (e.g., a frown, a head shake, or vocal response such as “No” or “Bad”) that typically precedes withholding of a reinforcer will only function as an effective conditioned punisher when the MOs for the withheld reinforcers are in effect. However, it is possible for a stimulus to be associated with several different punishers. In this case, this stimulus may function as a generalized conditioned punisher, whose effectiveness will be almost free from the influence of MOs. If the disapproving look has been associated with the loss of a wide range of reinforcers, including generalized reinforcers such as money, then the look may almost always function as a punisher, since at least one of the reinforcers will be valuable at any given moment. The same applies for the negative punishment procedures time-out and response cost (see Chapter 15). Time-out will only be effective when “time-in” serves as a reinforcer when the behavior is being punished. Solnick, Rincover, and Peterson (1977) compared the effectiveness of time-out when the time-in environment was either impoverished or enriched when trying to decrease self-injury and stereotypy in a child diagnosed with intellectual disabilities. The impoverished time-in condition looked no different from the child’s typical treatment session, in which correct responses (sorting blocks) were reinforced with praise, food, and toys. During the enriched time-in condition, the experimenters played music, introduced new toys, and increased social interaction by continuously prompting the participant to play with the toys. Time-out consisted of 90 sec of no access to reinforcers contingent upon problem behavior. Their results showed that time-out was effective when the time-in was enriched and generally ineffective when time-in was impoverished. This suggests that the effectiveness of a negative punishment procedure depends upon the MO for the reinforcer being eliminated. The same can be said about response cost; unless the events removed contingent upon behavior are effective as reinforcers at the time the procedure is implemented, response cost will not serve as an effective punishment procedure. In general, when observing a punishment effect, one must also consider the status of the reinforcer responsible for the occurrence of the punished behavior. A behavior is targeted for reduction only if it has been occurring and its occurrence is a function of reinforcement. For instance, disruptive behavior maintained by attention may increase in frequency when it has not received much attention (MO for reinforcement). Any punishment procedure targeting this disruptive behavior will have to be delivered when the MO for attention is in effect. Suppose a time-out procedure had been used to punish disruptive behavior. As mentioned earlier, only when the MO relevant to the reinforcer available (attention during time-in) was in effect should we expect time-out to function as punishment. And then only if those MOs were in effect should we expect to see the abative effect of the punishment procedure on disruptive behavior (i.e., immediate reduction in behavior). In the absence of the MO for disruptive behavior (lack of attention), there would be no behavior to be abated. Although these complex behavioral relations have not received much attention in the conceptual, experimental, or applied behavior analytic literatures, they seem to follow naturally from existing knowledge of reinforcement, punishment, and motivating operations. Behavior analysts should be aware of the possible participation of these behavioral relations in any situation involving punishment.11 MULTIPLE EFFECTS OF MOs Most environmental events have more than one effect upon behavior. It is important to recognize these various effects and to not confuse one effect with another (Skinner, 1953, pp. 204–224). Multiple effects are readily apparent in an animal laboratory demonstration of an operant chain. A food-deprived rat pulls a cord hanging from the ceiling of the chamber. Pulling the cord produces an auditory stimulus such as a buzzer. In the presence of the buzzer, the rat presses a lever and receives a food pellet. The onset of the buzzer has two obvious effects: (1) It evokes the lever press, and (2) it increases the future frequency of cord pulling. The buzzer functions as an SD for the evocative effect and as a conditioned reinforcer for the reinforcement effect. Both effects influence behavior in the same direction: an increase in the current frequency and an increase in future 11It is sometimes argued that behavioral principles are too simplistic for analyzing complex human behavior, and therefore some nonbehavioral—usually cognitive—approach is required. It may be that the behavioral repertoire of the people who make this argument is too simple for the task. However, behavioral principles themselves are not so simple, as can be seen from the preceding effort to understand the effects of punishment MOs, or the learned functions of painful stimulation. frequency of the lever press.12 Similarly, a hungry person may open the fridge to find her roommate’s leftover Mexican food. The sight of the Mexican food evokes behaviors associated with eating it (e.g., getting it out of the fridge, heating it up), as well as possibly increasing the future frequency of opening the fridge. Similarly, if pressing the lever in the presence of the buzzer produces an electric shock, then the sound of the buzzer would function as a discriminative stimulus for punishment (SDp ). The buzzer would then have abative effects on the current frequency of lever pressing and function as a conditioned punisher that decreases the future frequency of cord pulling. Likewise, if the leftover food is spoiled, then it abates the behavior of eating it as well as decreases the behavior of opening the fridge to look for food. Like SDs, events that function as UMOs will typically have behavior-altering effects on the current frequency of a type of behavior and, when presented as consequences, will have function-altering effects on the future frequency of behavior that immediately preceded their onset. An increase in painful stimulation will, as an antecedent, have MO establishing effects that would increase the current frequency of all behavior that has alleviated pain and, as a consequence, would decrease the future frequency of whatever behavior immediately preceded the increased pain. As in the example above, events that have a UMO evocative effect, such as painful stimulation, will also function as punishment for the response immediately preceding the onset of the event. However, this statement must be qualified for events that have such gradual onsets (like food or water deprivation) that they cannot easily function as consequences. In other words, even though food deprivation can evoke all behaviors that have produced food in the past, it would likely not be effective as a punishing consequence to decrease the future frequency of behavior. Events that have MO abolishing effects as antecedents may have reinforcing effects when presented as consequences for behavior. For example, food consumption momentarily decreases the reinforcing value of food as an effective reinforcer (abolishing effect), as well as momentarily decreases behaviors that have produced food in the past (abative effect). As a consequence, food consumption increases the future frequency of behaviors that have produced it (reinforcing effect). Thus, it is important to note that when edibles are used as consequences for increasing behavior (reinforcement effect), the repeated consumption of edibles decreases their value as an effective reinforcer (abative effect). Similarly, if an item is removed in an attempt to increase its value as a reinforcer (establishing effect), its removal may reduce the future frequency of whatever behavior immediately preceded the removal (punishing effect). Tables 16.3 and 16.4 illustrate the multiple effects of environmental events.

CONDITIONED MOTIVATING OPERATIONS (CMOs) Motivating variables that alter the reinforcing effectiveness of other stimuli, objects, or events as a result of the organism’s learning history are called conditioned motivating operations (CMOs). As with UMOs, CMOs alter the momentary frequency of all behavior that has been reinforced by those other events. In commonsense terms, some environmental variables, as a product of our experiences, make us want something different from what we wanted prior to encountering those variables, and encourage us to try to obtain what we now want. For example, needing to enter a room with a locked door establishes the key to the lock as an effective reinforcer. A locked door is a CMO because its value-altering effect is a function of a learning history involving doors, locks, and keys. There seem to be at least three kinds of CMOs, all of which are motivationally neutral events prior to their relation to another MO or to a form of reinforcement or punishment. The distinction among these conditioned motivating operations is based upon how they were developed and how they affect behavior. A surrogate conditioned motivating operation (CMO) produces the same effects as another MO because of some type of pairing (i.e., a surrogate is a substitute or stand-in), a reflexive CMO alters a relation to itself (makes its own removal effective as reinforcement), and a transitive CMO makes something else effective as reinforcement (rather than altering itself). Surrogate CMO (CMO-S) A CMO-S is a previously neutral stimulus that acquired its MO effects by being paired with a UMO. This process is similar to how respondent conditioned stimuli, operant conditioned reinforcers, and operant conditioned punishers acquire their functions by being paired with another behaviorally effective stimulus. For example, a restaurant may exert discriminative control over several behaviors, including ordering food, since restaurants have been differentially correlated with the availability of food. However, because we typically go to restaurants after a period in which we have not had any food (i.e., food deprivation as a UMO), it is possible that after multiple pairings with food deprivation, the restaurant would function as a CMO-S, (a) establishing food as a reinforcer and (b) evoking behaviors that have produced food (e.g., ordering some). Figure 16.6 exemplifies the establishment of the restaurant as a CMO-S. However, the existence of this type of CMO seems to be opposed to the organism’s best interest for survival (Mineka, 1975). For example, ordering food when satiated, as in the example above, would not seem healthy. However, a number of animal studies have shown that a stimulus paired with a reinforcer (e.g., food) under a high (e.g., 24 h) deprivation condition is preferred over a stimulus that was paired with the same reinforcer amount under a low (e.g., 12 h) deprivation condition (e.g., Lewon & Hayes, 2014; Vasconcelos & Urcuioli, 2008). In other words, when two stimuli are paired with the same reinforcer, but under different MO conditions, subjects that are mildly deprived chose the stimulus correlated with highdeprivation levels. It is possible that in addition to the food deprivation (mild deprivation), the presence of the stimulus correlated with high deprivation acts as a CMO-S, which makes the subjects behave as if they are “hungrier” than they would be otherwise, and this might be why they tend to respond more frequently to the high-deprivation alternative. Durlach, Elliman, and Rogers (2002) have found that adults under the same MO condition chose a flavored and colored drink that had been correlated with a high-value UMO for liquids (high-salt meal) over a drink that had been paired with a low-value UMO (low-salt meal), suggesting that the flavor and color acquired CMO-S properties. O’Reilly, Lancioli, King, Lally, and Dhomhnaill (2000) found that aberrant behavior (pushing, pinching, property destruction, and self-injury) displayed by two individuals with developmental disabilities was maintained by attention when their parents were interacting with a third person (diverted attention). Given that parents provided low levels of attention when they were talking to a third person, diverted attention was paired with low attention (MO). This may have led to diverted attention functioning as a CMO-S. Figure 16.7 shows that for both participants, problem behavior that never occurred during the brief functional analysis did so when the diverted attention condition (CMO-S) was introduced. Providing attention noncontingently (abolishing operation) produced significant reductions in problem behavior. Follow-up observations showed reduced levels of problem behavior. These results seem to suggest that previously neutral stimuli may acquire MO functions due to being paired with another effective MO. In a translational evaluation of the CMO-S, Lanovaz, Rapp, Long, Richling, and Carroll (2014) repeatedly paired a poster board with the presence of a condition (last 2 minutes of each session) known to function as an EO for stereotypy. After this pairing procedure, three out of four participants diagnosed with autism displayed higher rates of stereotypy in the presence of the poster board. These results have important applied implications, as there may be many environmental features in a therapeutic setting that could acquire CMO-S control over problem behavior simply by being paired with another MO. If a toy known for evoking stereotypy is delivered after compliance for a specific instruction, this repeated pairing may establish the instruction as a CMO-S that would increase the value of a specific form of automatic reinforcement and evoke stereotypy (Lanovaz et al., 2014). The relationship between the CMO-S and an effective MO can generally be weakened by two kinds of MO unpairing: presenting the previously neutral stimulus (now CMO-S) without the effective MO or presenting the MO as often in the absence of the CMO-S as in its presence. In the examples above, if the instruction is no longer followed by the toy that usually evokes stereotypy, or a parent talking to a third person is no longer followed by low levels of attention, the value-altering and behavioraltering effects of the toy or diverted parental attention would be weakened. Similarly, if the toy known to evoke stereotypy is presented as often in the presence of the instruction as in its absence, or if differing levels of attention are associated with talking to a third person, their motivating effectiveness would be reduced. The CMO-S appears to play a role in the development of behaviors that do not seem to make much sense, such as seeing a restaurant after dinner and wanting to eat again or feeling the urge to urinate after seeing the sign for the restroom. Both the restaurant and the restroom may evoke behavior as a CMO-S, given previous pairings with food deprivation and a full bladder, respectively (Miguel, 2013). Reflexive CMO (CMO-R) Any stimulus that systematically precedes the onset of painful stimulation becomes a CMO-R, in that its own offset functions as a reinforcer, and its occurrence evokes any behavior that has produced such reinforcement in the past. For example, in a traditional discriminated avoidance procedure,13 after some interval of time, the onset of an initially neutral stimulus is followed by the onset of painful stimulation—usually electric shock (see Box 16.3, “What’s an Aversive Stimulus?”). Some arbitrary response (one that is not part of the organism’s phylogenic repertoire), such as lever pressing, terminates the painful stimulation (the animal escapes the pain) and restarts the interval. If the same response occurs during the warning stimulus, it terminates that stimulus and the shock does not occur on that trial. The response at this phase of the procedure is said to have avoided the pain and is called an avoidance response. Many organisms learn to respond during the onset of a warning stimulus, and thus receive very few shocks. In this case, the warning stimulus functions similarly to the shock as an MO for the escape response, the reinforcer for which is shock termination. However, the capacity for the warning stimulus to establish its own termination as an effective reinforcer is a result of the organism’s learning history, involving the pairing of the warning stimulus and shock. In other words, the warning stimulus evokes the avoidance response as a CMO, just as the painful stimulation evokes the escape response as a UMO. The warning stimulus is not an SD correlated with the availability of the consequence, but rather an MO related to the reinforcing effectiveness of the consequence. Recall that a discriminative stimulus is related to the current availability of a type of consequence for a given type of behavior. Availability is defined by two components: (a) An effective consequence (i.e., one whose MO is currently in effect) must have followed the response in the presence of the stimulus; and (b) the response must have occurred without the consequence (which would have been effective as a reinforcer if it had been obtained) in the absence of the stimulus. The relation between the warning stimulus and consequence availability does not meet the second requirement. In the absence of the warning stimulus, there is 13The term discriminated avoidance arose so that this type of procedure could be distinguished from an avoidance procedure in which an exteroceptive stimulus change does not precede the shock itself (called avoidance without a warning stimulus, non-discriminated avoidance, or free-operant avoidance). See Chapter 12. no effective consequence that could have failed to follow the response as an analog to the extinction responding that occurs in the absence of an SD. The fact that the avoidance response does not turn off the absent warning stimulus is like the unavailability of food reinforcement for a food-satiated organism. By contrast, a stimulus that has preceded some form of improvement may function as a CMO-R by (a) establishing its own offset as an effective punisher and (b) abating any behavior that has been so punished. For example, a positive evaluation from an employer may have been followed by a promotion or some sort of monetary bonus. This history would establish the offset of positive evaluations as a form of punishment and likely reduce the frequency of all behaviors that may have resulted in a bad evaluation (e.g., being late). Although this relation is quite plausible, little research has seemed to focus on this type of CMO-R. The CMO-R plays an important role in identifying the negative aspects of many everyday interactions. Imagine that a stranger asks you where a particular building is located or asks for the time. The appropriate response is to give the information quickly or to say you don’t know. Typically, the person who asked will smile and say, “thank you”. Also, your question may be reinforced by the knowledge that a fellow human being has been helped (Skinner, 1957). In a sense, the question is an opportunity to obtain reinforcers that were not previously available. However, the question also begins a brief period that can be considered a warning stimulus, and if a response is not made soon, a form of social worsening (awkwardness) will occur. The stranger may repeat the question, stating it more clearly or more loudly, and will certainly think you are socially inept if you do not respond quickly. You would also consider it inappropriate to provide no answer. Even when no clear threat for nonresponding is implied by the person who asked, our social history under such conditions implies a form of worsening for continued inappropriate behavior. Many such situations probably involve a mixture of positive and negative components, but in those cases in which answering the question is an inconvenience (e.g., the listener is in a hurry), the stranger’s thanks is not a strong reinforcer, nor is helping one’s fellow human being, and the CMO-R is probably the main controlling variable. Another example is when a person thanks another for doing a kindness of some sort. Saying “thanks” is evoked by the person’s performing the favor or the kindness. Performing the favor may be considered an SD in the presence of which one has learned to say “Thank you” and received reinforcement consisting of the other person saying, “You’re welcome.” However, in many cases, ordinary courteous remarks may involve a CMO-R component. Consider the following scenario. Bill has his arms full, carrying something out of the building to his car. As he approaches the outer door, Danielle opens the door and holds it for Bill. Ordinarily, Bill would then smile and say, “Thanks.” The CMO-R component can be illustrated by supposing that Bill just walks out without acknowledging the favor. In such circumstances, it would not be unusual for Danielle to call out sarcastically, “You’re welcome!” Someone’s favor may function as a warning stimulus (a CMO-R) that has systematically preceded some form of disapproval in the absence of some form of acknowledgement (saying “Thanks”). In the typical laboratory avoidance procedure, the response terminates the warning stimulus. In extending this analysis to the human situation, it must be recognized that the warning stimulus is not simply the event that initiated the interaction. In the previous example, the CMO-R is not the vocal request itself, which is too brief to be terminated. It is, instead, the more complex social stimulus situation consisting of having been asked and not having made a response during the time when such a response would be appropriate. The termination of that situation is the reinforcement for the response. Some social interactions with a stimulus—a facial expression, an aggressive posture—are more like the warning stimulus seen in the laboratory that can be terminated by the avoidance response, but most involve the more complex stimulus condition consisting of the request and the subsequent period described earlier. In applied behavior analysis, the CMO-R is a component of training procedures for teaching individuals with defective social and verbal repertoires. During early intensive behavioral intervention, for example, learners are typically asked questions or given verbal instructions, followed by verbal or physical prompts so they can respond correctly. This arrangement may function as CMO-Rs, which will be followed by further intense social interaction if they do not respond appropriately (e.g., error correction). Such questions and instructions may be functioning primarily as CMO-Rs, rather than SDs related to the possibility of receiving praise or other positive reinforcers (Carbone et al., 2010). In other words, in these situations, it is likely that correct responses are partially maintained by negative rather than positive reinforcement. In addition to establishing their own removal as a reinforcer and evoking escape-maintained behaviors, these instructions may establish the value of activities or items unrelated to the task at hand as conditioned reinforcers. In this case, the instructions would function as transitive CMOs, as described in the following section (Michael, 2000). Figure 16.8 illustrates how, after preceding an aversive event, an instruction may become a CMO-R that would (a) establish its own removal as a conditioned reinforcer and (b) evoke any behavior that has produced such a reinforcer, such as compliance with the instruction. Behaviors maintained by negative reinforcement are under the influence of a variety of antecedent stimuli, including task difficulty as a CMO-R. Changing these tasks may abolish the reinforcing effects of negative reinforcement and abate undesirable escape behaviors (Smith & Iwata, 1997). As an attempt to assess the function of problem behavior, as well as identify the specific features of the environment that served as EOs, McComas, Hoch, Paone, and El-Roy (2000) exposed three children diagnosed with developmental disabilities to a typical experimental functional analysis, a descriptive assessment to generate hypotheses about possible antecedents for destructive behavior, and an analysis in which they tested the evocative effects of the hypothesized EOs. Although all participants engaged in problem behavior in the presence of task demands, the specific task features that evoked problem behavior (CMO-Rs) varied across participants. However, problem behavior by all participants was reduced simply by making specific modifications to the instructional methods, without modifying the demands. For one participant (Eli), problem behavior was evoked by a novel or difficult task (CMO-R). During the EO analysis, the authors compared the level of problem behavior when an instructional strategy was used (i.e., manipulables available during a math task) versus when the strategy was not used. Problem behavior rarely occurred when the strategy was used, suggesting that the establishing and evocative effects of task demands on escape-maintained problem behavior were eliminated by the addition of the instructional strategy. This effect may have been achieved by making the tasks less difficult and, thus, unpairing task demands with aversive properties. Results from a sample participant are shown in Figure 16.9. Functional analysis (top panel) suggested that problem behavior consisted of escape from academic tasks. During the EO analysis and follow-up (middle panel), problem behavior did not occur when the instructional strategy was employed, compared to when it was not. Additionally, the participant was more compliant when the instructional strategy was available (bottom panel) Numerous treatments to weaken or abolish CMO-Rs that evoke problem behavior in the presence of task demands have been evaluated (see Carbone et al., 2010, for a review). These include programming competing reinforcers for compliance (e.g., Piazza et al., 1997), pairing stimuli that evoke problem behavior with reinforcement by embedding preferred activities during task presentation (e.g., Kemp & Carr, 1995), reducing the aversiveness of the task by using errorless procedures (Heckaman, Alber, Hooper, & Heward, 1998), systematically fading instructional demands (Pace, Iwata, Cowdery, Andree, & McIntyre, 1993), varying the tasks being presented (McComas et al. 2000; Winterling, Dunlap, & O’Neill, 1987), increasing the pace of instruction to produce higher rates of reinforcement for appropriate behavior (Roxburgh & Carbone, 2013), presenting highly preferred activities and low demands to neutralize CMO-Rs occurring prior to the session (e.g., Horner, Day, & Day, 1997), reducing the aversiveness of tasks by providing choices of activities and reinforcers during sessions (Dyer, Dunlap, & Winterling, 1990), interspersing easy tasks with difficult ones (Mace & Belfiore, 1990), gradually introducing novel tasks (Smith, Iwata, Goh, & Shore, 1995), and modifying task duration based on whether participants engage in problem behavior early or late in the session (Smith et al., 1995). It may also be possible to weaken the CMO-R by unpairing it with the aversive stimulus it precedes (Kettering, Neef, Kelley, & Heward, 2018). It is important to note that negatively reinforced problem behaviors improve environmental conditions that are otherwise poor. Thus, it is the behavior analyst’s ethical duty to identify and modify these aversive variables (including CMO-Rs), rather than simply reduce problem behavior through some form of consequence manipulation such as escape extinction (Carbone et al., 2010; McGill, 1999). Transitive CMO (CMO-T) When an environmental variable establishes the effectiveness of another event as a reinforcer or punisher, it is referred to as a CMO-T. This type of MO was previously described in the behavior analytic literature as an establishing stimulus (SE; Michael, 1982) or blocked-response CEO (Michael, 1988; also see Miguel, 2013). Conditioned reinforcers are dependent upon a variable functioning as a CMO-T. Thus, all variables that function as UMOs also function as CMO-Ts for the stimuli that are conditioned reinforcers because of their relation to the relevant unconditioned reinforcer. For example, not only is food deprivation a UMO that momentarily (a) establishes the value of food as an unconditioned reinforcer and (b) evokes all behaviors that have produced food in the past; food deprivation is also a CMO-T that momentarily (a) establishes the value of all conditioned reinforcers paired with food (e.g., an attentive server in a restaurant, a menu, utensils) and (b) evokes all behaviors that have produced these conditioned reinforcers in the past. Consider the simple operant chain described earlier: A food-deprived rat pulls a cord that turns on a buzzer. In the presence of the buzzer sound, the rat then presses a lever and receives a food pellet. Food deprivation as a UMO makes food effective as an unconditioned reinforcer, a relation that requires no learning history. Food deprivation as a CMO-T also makes the buzzer sound effective as a conditioned reinforcer, which clearly requires a learning history. Thus, food deprivation is a UMO with respect to the reinforcing effectiveness of food, and a CMO-T with respect to the reinforcing effectiveness of the buzzer sound. Conversely, food ingestion is a UMO that abolishes the reinforcing effectiveness of food and the buzzer sound. In the Mexican restaurant, we may need to call the waiter to get his attention prior to ordering our tacos. In this scenario, food deprivation functions as a UMO that establishes food as an unconditioned reinforcer, and as a CMO-T that establishes the sight of the waiter as a conditioned reinforcer. As in the rat example above, food ingestion would decrease the reinforcing values of both food and the waiter. Other UMOs, such as painful stimulation, also establish the reinforcing value of conditioned reinforcers. As mentioned above, as a UMO, a headache would establish its own removal as an unconditioned reinforcer. However, as a CMO-T, a headache also establishes the reinforcing value of other stimuli (e.g., aspirin) that have served to reduce headaches. The relationship between UMOs and conditioned reinforcers cannot be overlooked by clinicians or educators who may make extensive use of conditioned reinforcers, including generalized ones, to strengthen academic or other socially significant behaviors. Studies have shown that the value of generalized conditioned reinforcers in the form of tokens decreases when participants have access to the primary and conditioned reinforcers paired with the tokens (Ivy, Neef, Meindl, & Miller, 2016; Moher, Gould, Hegg, & Mahoney, 2008). Conversely, when participants have restricted access to back-up reinforcers, tokens are more effective reinforcers, as evidenced by an increased rate of responding. In practice, these data suggest that the value of conditioned and generalized conditioned reinforcers (e.g., tokens) depends upon the MO relevant to their back-up reinforcers. Thus, it seems reasonable to assume that the effectiveness of generalized reinforcers would be increased when paired with a greater number of back-up reinforcers (Skinner, 1957, p. 54). More specifically, satiation may be prevented by pairing the putative generalized reinforcer with a variety of back-up reinforcers, such as food and drink that are under control of dissimilar MOs (Moher et al., 2008). Money, for instance, seems almost completely free from MO control, given that it has been paired with a multitude of effective reinforcers related to several different MOs. Thus, it is likely that one or more of these MOs will be in effect when money is delivered contingent on behavior, making its effectiveness almost impossible to abolish. The reinforcing effectiveness of many (probably most) conditioned reinforcers is not only altered by relevant UMOs, as described above, but also dependent on other stimulus conditions. This is why the effectiveness of conditioned reinforcers is often said to be “context” dependent. When the context is not appropriate, the stimuli may be available but are not accessed because they are not effective as reinforcers. A change to an appropriate context will evoke behavior that has been followed by those stimuli, which are now effective as conditioned reinforcers. The occurrence of the behavior is not related to the availability, but rather to the value, of its consequence. For example, flashlights are usually available in home settings but are not accessed until a power failure makes them valuable. In this sense, the power failure (the sudden darkness) evokes the behavior that has obtained the flashlight in the past (rummaging around in a particular drawer). However, when the light is back on, the value of the flashlight as a reinforcer is abolished, and any behaviors that led to the flashlight are momentarily abated. The motivating nature of this CMO-T relation is not widely recognized, and the evocative variable (the sudden darkness) in this example is usually interpreted as an SD. Consider a workman disassembling a piece of equipment, with his assistant handing him tools as he requests them.15 The workman encounters a slotted screw that must be removed and requests a screwdriver. The sight of the screw evoked the request, the reinforcement for which is receiving the tool. Prior to an analysis in terms of the CMO-T, the sight of the screw would have been considered an SD for the request, but such screws have not been differentially related to the availability of reinforcement for requests for tools. In the typical workman’s history, assistants have provided requested tools irrespective of the stimulus conditions in which the request occurred. The sight of the screw is more accurately interpreted as a CMO-T for the request, not as an SD. The fact that several SDs are involved in this complex situation makes the analysis somewhat difficult. The screw is an SD for unscrewing movements (with a screwdriver in hand) or for selecting a screwdriver rather than another tool. The workman’s verbal request for a screwdriver, although evoked by the sight of the slotted screw as a CMO-T, is dependent on the assistant’s presence as an SD. The offered screwdriver is also an SD for the workman reaching to grasp it. The critical issue here, however, is the screw's role in evoking the request, and this is a motivating rather than a discriminative relation (Michael, 1982). Another common example involves a stimulus related to some form of danger that evokes some relevant protective behavior. Imagine a night security guard is patrolling an area and hears a suspicious sound. He pushes a button on his phone that signals another security guard, who then activates his phone and asks if help is needed (which reinforces the first guard’s call). The suspicious sound is not an SD in the presence of which the second security guard’s response is more available (the second guard would respond to the first guard’s call whether or not the first guard heard the sound), but rather a CMO-T in the presence of which second guard’s response is more valuable. SDs are involved, however. A ringing phone is an SD in the presence of which one has activated the phone, said something into the receiver, and been reinforced by hearing a response from another person. Answering phones that are not ringing has typically not been so reinforced.16 (Note, incidentally, that the effect of the danger signal is not to evoke behavior that produces its own termination, as with the CMO-R, but rather behavior that produces some other event, in this case, the sound of the security guard’s colleague offering to help.) Figure 16.10 shows an example of sudden darkness functioning as a CMO-T that establishes a flashlight as a conditioned reinforcer, as well as evoking the behavior of looking for a flashlight. Note that sudden darkness may also function as a CMO-R that establishes its own removal as a reinforcer.17 The distinction between an SD and a CMO-T hinges on the relation between reinforcer availability and the presence or absence of the stimulus. If the reinforcer is more available in the presence than in the absence of the stimulus, the stimulus is an SD; if the reinforcer is equally available in the absence and presence of the stimulus, the stimulus is a CMO-T. Screwdrivers have typically been available to the workman in the absence as well as in the presence of screws. The response by the security guard’s colleague has been just as available in the absence as in the presence of a suspicious noise. A CMO-T evokes behavior because of its relation to the value of a consequence rather than to the availability of a consequence. The two forms of behavioral control, the SD and the CMO-T, which are so different in origin, would be expected to differ in other important ways. This is an example of terminological refinement in behavior analysis, not a discovery of new empirical relations. The value of this refinement will be found in the improved theoretical and practical effectiveness of behavior analysts whose verbal behavior has been affected by it. Teaching Mands with CMO-Ts. Mand training is an essential aspect of language programs for individuals with severely deficient verbal repertoires (see Chapter 18). For such individuals, manding does not spontaneously arise from tact and receptive discrimination training (i.e., listener behavior). The learner has to want something, make an appropriate verbal response, and be 16See Chapter 2, Box 2.2, “When the Phone Rings.” 17If removal of darkness is to be considered an unconditioned reinforcer, then sudden darkness would establish the value of its removal as a UMO. reinforced by receiving what was wanted. This procedure places the response under control of the relevant MO. The manipulation of CMO-Ts allows the teacher to contrive a situation to get the learner to want something that can be a means to another end (e.g., obtaining information by asking a question). Any stimulus, object, or event can be a basis for a mand simply by arranging an environment in which that stimulus can function as a conditioned reinforcer. Thus, if a pencil mark on a piece of paper is required for an opportunity to play with a favorite toy, a mand for a pencil and for a piece of paper can be taught. The manipulation of CMO-Ts is common practice when teaching mands to individuals with disabilities (e.g., Alwell, Hunt, Goetz, & Sailor, 1989; Betz, Higbee, & Pollard, 2010; Goetz, Gee, & Sailor, 1985; Lechago, Carr, Grow, Love, & Almason, 2010; Shillingsburg, Valentino, Bowen, Bradley, & Zavatkay, 2011). The typical procedure, termed interrupted-chain (Hall & Sundberg, 1987) involves withholding an item the learner needs to complete a previously learned task. This manipulation functions as a CMO-T, increasing the value of the item as a reinforcer, which creates the ideal condition for teaching the learner to mand for it. In a mand training study, Shillingsburg, Bowen, Valentino, and Pierce (2014) exposed three children with disabilities to two scenarios. One scenario involved varying the placement of a preferred item in one of nine different colored (or animal) cups. A stimulus that signaled the preferred item (e.g., an empty Skittles bag) was then placed next to the participant. When the participant asked for the item (e.g., “May I have a Skittle?”), the experimenter responded by saying that the item was under one of the cups, increasing the value of the information for the location of the item (which cup) as a conditioned reinforcer. The second scenario involved giving a highly preferred item to one of three adults when the participant was not looking. The stimulus that signaled the presence of the preferred item was presented again, and when the child asked for it, the therapist said that one of the other adults had it, without saying whom. This statement served as a CMO-T that increased the value of the information about who had the item as a conditioned reinforcer. The authors used a vocal prompt delay to teach participants to ask, “which cup?” and “who has it?” in scenarios 1 and 2, respectively. These CMO-T trials were alternated with trials in which the information was not valued as a reinforcer (abolishing operation; AO) to guarantee that behavior came under control of the EO, versus some other aspect of the environment. During AO trials, the therapist provided either the information about the location of the item or the name of the person who had it. In addition to the emission of mands, the authors also collected data on whether participants selected the correct cup or approached the correct adult to obtain the preferred item. All participants manded using “which” and “who” more often during EO rather than AO conditions, and approached the preferred items during both conditions, suggesting that the interrupted chain procedure was successful in placing mands under functional control of CMO-Ts. Figure 16.11 shows the results for the “which” questions. None of the participants manded for information during baseline (left panel), although each selected the correct container during the AO condition during which the information on the location of the item was provided (right panel). After training, participants manded only during the EO condition (left panel). Additionally, after receiving the information during the EO condition, participants started to select the correct containers to access the preferred item (right panel). Similar results were obtained for the “who” questions. Chapter 18 provides additional details about using CMO-Ts during mand training. RELEVANCE OF MOs TO THE GENERALITY OF TREATMENT EFFECTS In the applied area, the reinforcer-establishing effect of MOs seems to be well understood, as clinicians may temporarily withhold edibles, tangibles, and even social attention to make them more effective as reinforcers for behaviors being taught Figure 16.11 Example of control by a motivating operation a transitive CMO (CMO-T). From “Mands for Information Using “Who?” and “Which?” in the Presence of Establishing and Abolishing Operations” by M, A. Shillingsburg, C. N. Bowen, A. L. Valentino, and L. W. Pierce, 2014, Reproduced with permission of John Wiley & Sons Inc. 10 EO EO Baseline Posttraining Baseline Posttraining AO AO 5 0 0 2 4 6 8 10 12 14 16 18 20 22 24 Ian 10 5 0 0 2 4 6 8 10 12 14 16 18 20 22 24 Jeb 10 5 0 0 Figure 1. Cumulative record of independent responses for manding “which?” (left) and cumulative record for approach behavior (right) during baseline and posttraining probes for lan, Jeb, and Jen. 2 4 6 8 10 Trials Cumulative Mand - Which Cumulative Approach - Which Trials 12 14 16 18 20 22 24 Jen 10 5 0 0 2 4 6 8 10 12 14 16 18 20 22 24 Ian 10 5 0 0 2 4 6 8 10 12 14 16 18 20 22 24 Jeb 10 5 0 0 2 4 6 8 10 12 14 16 18 20 22 24 Jen (e.g., Taylor et al., 2005). However, these target behaviors will not occur in future or novel circumstances, even if well learned and part of the learner’s repertoire, unless relevant MOs are in effect (Fragale et al., 2012; O’Reilly et al., 2012). Fragale and colleagues have shown that after teaching three children diagnosed with autism to mand for preferred items and testing for generalization across settings, the newly acquired mands occurred almost exclusively when items had been restricted prior to the session and seldom occurred when participants could access the items prior to sessions. These results show that even if well established, a behavior may not occur in the future in a new setting if the MO is absent. Thus, it is possible that generalization and maintenance failures are not only a function of novel (generalization) stimuli being dissimilar to those present during training, but also due to different MOs. Thus, when programming for generalization, clinicians must do so not only across different discriminative stimuli (Stokes & Baer, 1977) but also across different MOs (Miguel, 2017). In summary, for a newly acquired response to be generalized and maintained, its relevant MO must also be in effect. RELEVANCE OF MOs TO APPLIED BEHAVIOR ANALYSIS Applied behavior analytic research has shown the effects of MOs on a multitude of behaviors across a variety of conditions. MOs have affected the results of preference assessments (Gottschalk, Libby, & Graff, 2000; McAdam et al., 2005) and functional analyses (Fischer, Iwata, & Worsdell, 1997; Worsdell, Iwata, Conners, Kahng, & Thompson, 2000), engagement with activities (Klatt et al., 2000; Vollmer & Iwata, 1991), within-session responding (Roane, Lerman, Kelley, & Van Camp, 1999), social initiations (Taylor et al., 2005), the rate of acquisition of target behaviors (Ivy et al., 2016), and the occurrence of challenging behaviors (O’Reilly et al., 2007). Gottschalk and colleagues (2000), for instance, compared the results of preference assessments of four children with autism under three conditions. In the control condition, participants had access to four edible items in premeasured portions, 3 times, 24 hr prior to the assessment. In the satiation condition, the same access to premeasured portions was provided, but the participants had free access to one of the stimuli for 10 min prior to the assessment. Finally, in the deprivation condition, participants had the same premeasured portions, but were deprived of one stimulus for 48 hr prior to the assessment. Results showed that across all participants, approach responses were higher during deprivation conditions. These results were replicated with leisure items with both typically developing children and individuals with intellectual disabilities (McAdam et al., 2005) and suggest that results of preference assessments are highly dependent on current MOs. As an attempt to evaluate the effects of MOs on the results of functional analyses, Worsdell et al. (2000) exposed six individuals with profound intellectual disabilities who engaged in self-injury to an experimental functional analysis (FA), followed by another FA in which the MO and the reinforcer for problem behavior were present or absent. During the EO condition, the experimenter ignored all but problem behavior, while in the AO condition, the experimenter delivered noncontingent reinforcement. These conditions were associated with either the presence or the absence of continuous reinforcement. The authors reported high rates of problem behavior for all participants during the condition in which the EO and reinforcement were present, and almost no responding when both the EO and reinforcement were absent, or when the EO was present and reinforcement was absent. These results are consistent with the conceptualization that for behavior to occur, it must be followed by an effective reinforcer, and suggest that MOs may differentially affect the results of FAs. As previously discussed, in addition to affecting results of commonly used behavior analytic assessments, MOs have been shown to directly affect the rate of acquisition of novel behaviors reinforced with primary, conditioned, or generalized reinforcers (e.g., Ivy et al., 2016; Lang et al., 2009; Moher et al., 2008; O’Reilly et al., 2009), as well as the frequency of behaviors that have already been acquired (e.g., Klatt et al., 2000). Behavior analysts have also directly manipulated MOs to reduce problem behaviors (e.g., Rapp, Cook, McHugh, & Mann, 2017) and teach new skills (Lechago et al., 2010). Research on MOs falls within the three broad categories previously described by Iwata, Smith, and Michael (1991): demonstrating the effects of MOs on behavior, using MOs to clarify results of assessments, and manipulating MOs to increase or decrease behavior. This ever-growing literature has had much impact on our practice. It allowed behavior analysts to recognize that the three-term contingency is dependent upon a fourth variable that modulates the effectiveness of reinforcement, and hence the evocative strength of discriminative stimuli. A clear understanding of these motivational variables seems critical for interpreting assessment results and designing effective behavior analytic interventions. As we go about our daily lives, we may clearly identify how MOs influence our own behavior. Revisiting the anecdote about our obsession with Mexican food, we see that food deprivation as a UMO, or, more specifically, not having had Mexican food for some time, increases its value as a reinforcer, and the frequency of previous behaviors that have produced it, such as going to our favorite Mexican restaurant. Food deprivation as a CMO-T establishes the sight of the restaurant as a conditioned reinforcer, increasing any behavior that preceded it (e.g., driving there) and making us happy when we see it. When at the restaurant, the sight of tortilla chips may increase the value of salsa as a reinforcer, as they tend to covary (CMO-S). The spiciness of the salsa establishes its own removal as a reinforcer (CMOR), which makes us drink water or any other liquid (e.g., too many margaritas). When the chips are gone (CMO-T), we call the waiter to bring us more. And even if we have already eaten, the sight of the restaurant alone (CMO-S) may get us to eat more. Given the ubiquity of motivational variables in our lives, a complete understanding of human behavior demands an analysis of the critical roles of MOs.