Behavioral Momentum Theory Of Punishment

Punishment is a response-dependent environmental change that reduces the future probability of a response (Azrin & Holz, 1966). Punishment contingencies are frequent in the natural environment (Skinner, 1953); however, our understanding of punishment contains substantial gaps, and basic research on punishment has declined rapidly over the past 40 years (Baron, 1991; Lerman & Vorndran, 2002). Furthering the understanding punishment as a process would have not only basic and theoretical implications, but applied implications as well, as common treatments (e.g., response blocking, guided compliance) may reduce problematic behavior through the mechanism of punishment (Lerman & Vorndran, 2002).
Both parameters of the punishing stimulus and the
reinforcer are relevant to response suppression (Azrin, 1966; Baron, 1991). Typically, greater response suppression is found if the punishing stimulus is abruptly rather than gradually introduced (Azrin, 1959a; 1960b; Azrin, Holz, & Hake, 1963; Holz & Azrin, 1963). Additionally, greater intensities of the punishing stimulus (e.g., higher amperage in case of shock) generally produces greater response suppression than lower intensities (Azrin, 1959b; 1960a; Azrin et al., 1963; Brethower & Reynolds, 1962). In general, continuous (or fixed-ratio (FR) 1) schedules of punishment produce greater response suppression than do intermittent schedules (Azrin et al., 1963). The immediacy with respect to the target response (Azrin & Holz, 1966) and response-dependency (Azrin, 1956) are also important parameters of the punisher. Regarding parameters of the reinforcer, less response suppression occurs when the organism is more deprived of the reinforcer (Azrin et al., 1963). The baseline schedule of reinforcement also influences the suppressive effects (Baron, 1991). In general, FR1 reinforcement schedules are correlated with more response-suppression than are variable-interval (VI) schedules (Azrin & Holz, 1966; Baron, 1991). The effects of reinforcement rate on the efficacy of punishment, however, are perhaps less straightforward.
A review of early punishment literature reveals that response suppression was thought to be inversely related to increased reinforcement rate or amounts of a reinforcer (e.g., Bower & Miller, 1960; Estes, 1944, Exp. F; Miller,1960;). However, very few investigations have directly studied this relation. Church & Raymond (1967) trained one group of 24 rats to lever press on a VI 5-min schedule and a second group of 24 rats on a VI 1/5 min schedule. Following baseline, half of the rats from each group experienced punishment training. The experimental group experienced a VI 2-min schedule of punishment, such that, on the average of every 2 min., a shock (.15 mA) became available and occurred immediately following the next response. The other half of each group served as the unpunished control group and experienced conventional extinction. Church & Raymond (1967) found that the amount of response suppression was inversely related to the rate of positive reinforcement, in that the VI 5-min experimental group showed greater response suppression than did the VI 1/5-min experimental group. In a similar investigation, Holz (1968) first established responding on a concurrent VI 3-min VI 3-min baseline. Following, a concurrent VI 1.9-min VI 7.5-min was in effect so that the proportion of reinforcements were roughly 4:1 on the two keys. An FR 1 schedule of punishment was then introduced with a series of progressively increasing punisher intensities, ranging from of 3-12 milliamps (mA). Unlike the findings of Church & Raymond (1967), Holz (1968) found (proportional) response suppression to be independent of reinforcement rate. Upon secondary data analysis of other experimental data sets using similar punishment procedures (i.e., Azrin, 1959b; 1960a, Azrin & Holz, 1961; Holz, Azrin, and Ulrich, 1963), Holz (1968) found proportional response suppression to be very similar across investigations despite vast differences in reinforcement rates. Finally, Bouzas (1978) arranged a two-component multiple schedule with a VI 1-min in one component and a VI 4-minute in the other. Responding was then punished with a brief (.35s) electric shock of varying intensities (1-5 mA) according to a VI 30-s schedule in both components of the multiple schedule. Similar to the findings of Church and Raymond (1967), a greater suppressive effect was found on the responding maintained by the VI 4-min schedule. Therefore, there are conflicting findings suggesting that response suppression is either inversely related to (Bouzas, 1978; Church & Raymond, 1967) or independent of (e.g., Holz, 1968) reinforcement rate. Several limitations of the investigations further impede making strong conclusions. Aside from the different experimental designs and schedules of punishment used, reinforcement rate in each of these investigations was confounded by accompanying changes in response rate, making interpretation based solely on the contribution of reinforcement rate impossible.
Behavioral Momentum Theory and Resistance to Change Behavioral Momentum Theory proposes that response rate and resistance to change, or the persistence of a response following a disruptor, are two separate aspects of behavior (Nevin & Grace, 2000).
Resistance to change is thought to be dependent upon the Pavlovian stimulus-reinforcer relation in the three term contingency, and independent of the response-reinforcer relation (Nevin & Grace, 2000). General findings are that responses maintained on richer schedules of reinforcement are more resistant to change than those maintained on leaner schedules of reinforcement (Nevin & Grace, 2000). Resistance to change is often measured as responses in a particular session following disruption as proportion of an average of baseline, which accounts for differences in the baseline rates of responding, unlike absolute response rates (Nevin, Mandell, & Atak, 1983). The findings of Behavioral Momentum Theory have been consistent across a variety of disruptors, such as extinction (Nevin, 2012), pre-feeding (Nevin, 1974), and response-independent reinforcers provided during intercomponent-intervals (Harper,
1996).
However, resistance to change may not be entirely determined by stimulus-reinforcer relations, and may in fact not be independent of the response-reinforcer relation.
In a multiple schedule, Blackman (1968b; Exp. 2) arranged conditions with nearly identical reinforcement rates but employed pacing schedules such that response rates higher in on component (VI DRH) than the other component (VI DRL). Contrary to predictions of Behavioral Momentum Theory, rates in high rate component (VI DRH) were relatively more suppressed by the pre-shock stimulus than were low rates (VI DRL). Additional support for the contribution of the response-reinforcer relation was provided by Lattal (1989). Lattal (1989) arranged a two component multiple schedule, in which a tandem VI 100-s (or 300-s for two pigeons) FR 10 was in effect one component, and a tandem VI 100-s (or 300-s of for two pigeons) DRL 5-s schedule was in effect the other component. These conditions arranged different response rates but equated reinforcement rates. Baseline conditions then alternated with conditions in which variable time (VT) food was delivered during the 15-s blackout period between components. Results showed that response rates in the tandem VI DRL component changed less than those in the tandem VI FR component and therefore, lower responses rates were more resistant to change than were higher response rates when reinforcement was held constant. Similarly, Nevin, Grace, Holland, and McLeen (2001; Exp. 2)
evaluated resistance to change in a multiple VI, variable-ratio (VR) schedule by adjusting the VR value such that the reinforcement rate was equated across the components, but the VR component resulted in higher response rates than did the VI component. Responding in the component associated with the VI was found to be more resistant to change than was higher-rate VR responding, and this was true across a variety of disruptors (i.e., food presented during the intercomponent-interval, extinction, and food during the intercomponent interval plus extinction). These findings were consistent with the conclusions of Blackman (1968b; Exp. 2) and Lattal (1989).
In summary, when response rates and reinforcement rates are arranged independently of one another in multiple-schedule arrangements, following the introduction a variety of different disruptors, lower responses rates are more resistant to change than are higher response rates (Blackman, 1968b; Exp. 2; Lattal, 1989; Nevin et al., 2001; Exp. 2). Although the aforementioned investigations have all used multiple schedules, da Silva, Maxwell, and Lattal (2008) provides a procedure wherein one may arrange reinforcement rate and response rate independent of one another in a concurrent arrangement, allowing for the direct comparison of manipulations of simultaneously available alternatives. da Silva et al. (2008; Exp. 2) independently arranged concurrent tandem VI 27-s FR 5 tandem VI 27-s DRL 3-s schedules in baseline to arrange for high and low rates of responding with equated rates of reinforcement. Alternatively, da Silva et al., (2008; Exp. 3) used a concurrent tandem VI 30-s DRL x-s tandem VI 30-s DRH y/ 3-s schedule in baseline to arrange nearly equivalent response rates on each key, while reinforcement rates were arranged such that fewer reinforcers were obtained on the key correlated with the tandem VI DRH schedule than on the key correlated with the tandem VI DRL schedule. DRH and DRL values were adjusted prior to each session to obtain the desired reinforcer ratio between the two keys. The arrangements of da Silva et al. (2008; Exp. 2; Exp. 3) may be useful in further exploring the aforementioned discrepancies of the punishment literature, particularly in studying punishers in concurrent arrangements, similar to the design of Holz (1968), but in a more precise manner to parse apart the independent effects of reinforcement rates and response rates on response suppression. Additionally, if findings using punishment are consistent with other disruptors (e.g., extinction), this may allow for the convergence of these various literatures.
Statement of the Problem
Despite frequent occurrence in the natural environment, our basic understanding of the process of punishment contains substantial gaps, including understanding the interaction of response suppression and reinforcement rate. Several experiments have studied the effects of reinforcement rate on the suppressive effects of punishment (e.g., Church & Raymond, 1967; Holz, 1968; Bouzas, 1978), but all have confounded changes in reinforcement rate with changes in response rate. An inverse relation between reinforcement rate and resistance to change following disruption is supported by Behavioral Momentum Theory, which suggests that resistance to change is dependent upon the Pavlovian stimulus-reinforcer relation and independent of the response-reinforcer relation (Nevin & Grace, 2000). However, when responses rates are held constant, lower response rates are more resistant to change than higher response rates (Blackman, 1968b; Exp 2; Lattal, 1989; Nevin et al., 2001; Exp. 2 ). Punishment has received little attention in the resistance to change literature, although findings on resistance to change to pre-aversive stimuli (Blackman, 1968b; Exp. 2) are similar to other disruptors. The following experiments are designed to examine response suppression 1) under conditions in which reinforcement rates between two concurrently available alternatives differ and response rates are equated and 2) under conditions in which response rates between two concurrently available alternatives differ and reinforcement rates are equated.