正强化和负强化:概念、争议与神经机制

1975年,Michael首次正式提出应放弃"正强化"和"负强化"术语的使用,引起对"正强化和负强化是否本质上具有区别"这一问题的关注和争议。正负强化究竟本质上是否有区分、如何区分、以及区分它们的意义等问题,得到了众多行为心理学家的深入探讨。多数学者再度阐明正强化和负强化具有不同的心理机制,加以区分具有重要应用意义,主张沿用对它们的区分。最近,对正负强化神经机制的研究取得突破性进展,进一步揭示它们具有截然不同的神经相关过程。我们建议,分析强化作用进行时,机体是否在环境条件作用下已经偏离了基础、正常的心理稳态,有助于规范理解强化物启动的是正还是负强化的神经过程。     

The distinction between positive and negative reinforcement: use with care

It is customary in behavior analysis to distinguish between positive and negative reinforcement in terms of whether the reinforcing event involves onset or offset of a stimulus. In a previous article (Baron & Galizio, 2005), we concluded that a distinction of these terms is not only ambiguous but has little if any functional significance. Here, we respond to commentaries by a group of distinguished behavior analysts about the issues we raised. Although several of the commentators argued for preservation of the distinction, we remain unconvinced that its benefits outweigh its weaknesses. Because this distinction is so deeply embedded in the language of behavior analysis, we hardly expect that it will be abandoned. However, we hope that the terms positive and negative reinforcement will be used with circumspection and with full knowledge of the confusion they can engender.     

AVERSIVE CONTROL: A SEPARATE DOMAIN?

Traditionally, aversive control has been viewed as a separate domain within behavior theory. Sometimes this separateness has been based upon a distinction between reinforcement and punishment, and sometimes upon a distinction between positive and negative reinforcement. The latter is regarded here as the more compelling basis, due to some inherent procedural asymmetries. An approach to the interpretation of negative reinforcement is presented, with indication of types of experiments that support it and that also point to promising directions for further work. However, most of the interpretive issues that arise here are relevant to positively reinforced behavior as well. These include: possible reformulation of the operant/respondent distinction; the place of emotional concepts in behavior analysis; the need for simultaneous, complementary analysis on differing time scales; the understanding of behavioral situations with rewarding or aversive properties that depend as much upon the contingencies that the situations involve as upon the primary rewarding or aversive stimuli that they include. Thus, an adequate understanding of this domain, which has been traditionally viewed as distinct, has implications for all domains of behavior-analytic theory.     

Positive and negative reinforcement: Should the distinction be preserved?

Michael (1975) reviewed efforts to classify reinforcing events in terms of whether stimuli are added (positive reinforcement) or removed (negative reinforcement). He concluded that distinctions in these terms are confusing and ambiguous. Of necessity, adding a stimulus requires its previous absence and removing a stimulus its previous presence. Moreover, there is no good basis, either behavioral or physiological, that indicates the involvement of distinctly different processes, and on these grounds he proposed that the distinction be abandoned. Despite the cogency of Michael's analysis, the distinction between positive and negative reinforcement is still being taught. In this paper, we reconsider the issue from the perspective of 30 years. However, we could not find new evidence in contemporary research and theory that allows reliable classification of an event as a positive rather than a negative reinforcer. We conclude by reiterating Michael's admonitions about the conceptual confusion created by such a distinction.     

Defining reinforcess — A problem in communication for consultation in behavior modification

This paper reviews the respondent (Hull-Spence) and operant (Skinnerian) conditioning definitions of reinforcers and reinforcement and demonstrates the need to keep the systems separate when consulting about behavior modification. The two systems are shown to lead to different modification procedures.One important distinction between the systems is whether a reinforcer is simply associated with a response (respondent) or whether it must follow the response (operant). A second important distinction is the definition of negative reinforcement. In respondent conditioning, negative reinforcement entails presenting an aversive stimulus in association with the response and results in a decrease in response rate. In operant conditioning, negative reinforcement entails the removal of an aversive stimulus following a correct response, which results in an increase in response rate.     

Fixed-interval limited-hold avoidance with and without signalled reinforcement

Rats trained to lever press on a fixed-interval limited-hold avoidance schedule maintained a pattern of responding similar to that maintained by fixed-interval limited-hold schedules of positive reinforcement. But this positively accelerated pattern of behavior was maintained only when the occurrence of reinforcement was signalled by the presentation of a brief flash of light. This result suggests that the discriminative function of the reinforcer in avoidance is less pronounced than the discriminative function of the reinforcer in escape or positive reinforcement. It also suggests that the distinction between positive reinforcement and avoidance is not superfluous. Although the schedule of reinforcement is an important variable in determining the pattern of behavior, other variables, such as the nature (i.e., stimulus presentation, termination, or omission) of the reinforcer, are also potent determinants of behavior.     

The effects of variable-time delivery of food items and praise on problem behavior reinforced by escape

Prior research indicates that reinforcement of an appropriate response (e.g., compliance) can produce concomitant reductions in problem behavior reinforced by escape when problem behavior continues to produce negative reinforcement (e.g., Lalli et al., 1999). These effects may be due to a preference for positive over negative reinforcement or to positive reinforcement acting as an abolishing operation, rendering demands less aversive and escape from demands less effective as negative reinforcement. In the current investigation, we delivered a preferred food item and praise on a variable-time 15-s schedule while providing escape for problem behavior on a fixed-ratio 1 schedule in a demand condition for 3 participants with problem behavior maintained by negative reinforcement. Results for all 3 participants showed that variable-time delivery of preferred edible items reduced problem behavior even though escape continued to be available for these responses. These findings are discussed in the context of motivating operations.     

Analyzing behavior implied by EWA learning: An emphasis on distinguishing reinforcement from belief learning

An important issue in the field of learning is to what extent one can distinguish between behavior resulting from either belief or reinforcement learning. Previous research suggests that it is difficult or even impossible to distinguish belief from reinforcement learning: belief and reinforcement models often fit the empirical data equally well. However, previous research has been confined to specific games in specific settings. In the present study we derive predictions for behavior in games using the EWA learning model (e.g., Camerer & Ho, 1999), a model that includes belief learning and a specific type of reinforcement learning as special cases. We conclude that belief and reinforcement learning can be distinguished, even in 2×2 games. Maximum differentiation in behavior resulting from either belief or reinforcement learning is obtained in games with pure Nash equilibria with negative payoffs and at least one other strategy combination with only positive payoffs. Our results help researchers to identify games in which belief and reinforcement learning can be discerned easily.     

Perceived objectivity and the effects of evaluative reinforcement upon compliance and self-evaluation in blacks

Two experiments were conducted to investigate the effects of positive and negative reinforcements upon black college students. In the first experiment subjects received either positive or negative reinforcement from a black or a white Evaluator. Behavioral change (compliance) was greater in response to negative reinforcement from a black Evaluator than in response to negative reinforcement from a white Evaluator, while there were no divergent effects of Evaluator's Race upon compliance with positive reinforcement. The same pattern obtained in subjects' self-evaluations. The second experiment offered a test of the hypothesis that the objectivity of black Evaluators was perceived as greater than that of white Evaluators, and that a manipulation aimed at equating the Evaluators on objectivity would attenuate the effects of Evaluator's Race upon compliance and self-evaluation in response to negative reinforcement. These hypotheses were generally supported by the results.     

The effects of noncontingent reinforcement on task disengagement following failure of differential reinforcement treatment

Following the failure of applications of differential reinforcement (DR), the effects of reinforcement delivered noncontingently (NCR) on task disengagement and task completion were evaluated within reversal designs for three children with autism spectrum disorder. In Experiment 1, we compared DR and NCR schedules that combined positive and negative reinforcement. In Experiment 2, we replicated the results of Experiment 1 and then evaluated NCR schedules composed solely of negative reinforcement and NCR schedules that combined negative reinforcement with high-or low-preferred stimuli. In Experiment 3, we further analyzed NCR schedules that included different types and qualities of positive reinforcement, but without the inclusion of negative reinforcement. The results showed that NCR schedules using positive reinforcers can be effective as a treatment for task disengagement, but the type and preference of the reinforcer can alter effectiveness. These results are interpreted in terms of the motivating operations associated with task disengagement.     

Theoretical implications of quantitative properties of interval timing and probability estimation in mouse and rat

Although theoretical discussions typically assume that positive and negative reinforcement differ, the literature contains little unambiguous evidence that they produce differential behavioral effects. To test whether the two types of consequences control behavior differently, we pitted money‐gain positive reinforcement and money‐loss‐avoidance negative reinforcement, scheduled through identically programmed variable‐cycle schedules, against each other in concurrent schedules. Contingencies of response‐produced feedback, normally different in positive and negative reinforcement, were made symmetrical. Steeper matching slopes were produced compared to a baseline consisting of all positive reinforcement. This free‐operant differential outcomes effect supports the notion that that stimulus‐presentation positive reinforcement and stimulus‐elimination negative reinforcement are functionally “different.” However, a control experiment showed that the feedback asymmetry of more traditional positive and negative reinforcement schedules also is sufficient to create a “difference” when the type of consequence is held constant. We offer these findings as a small step in meeting the very large challenge of moving negative reinforcement theory beyond decades of relative quiescence.     

Development and modification of a response class via positive and negative reinforcement: a translational approach

When responses function to produce the same reinforcer, a response class exists. Researchers have examined response classes in applied settings; however, the challenges associated with conducting applied research on response class development have recently necessitated the development of an analogue response class model. To date, little research has examined response classes that are strengthened by negative reinforcement. The current investigation was designed to develop a laboratory model of a response class through positive reinforcement (i.e., points exchangeable for money) and through negative reinforcement (i.e., the avoidance of scheduled point losses) with 11 college students as participants and clicks as the operant. Results of both the positive and negative reinforcement evaluations showed that participants usually selected the least effortful response that produced points or the avoidance of point losses, respectively. The applied implications of the findings are discussed, along with the relevance of the present model to the study of punishment and resurgence.     

Reinforcement as a mediator of the perception of humans by horses (Equus caballus)

A central question in the interspecific human/animal relationship is how domestic animals perceive humans as a significant element of their environment. In this study, we tested the hypothesis that the use of positive or negative reinforcement in horse training may have consequences on the animals’ perception of humans, as a positive, negative or neutral element. Two groups of ponies were trained to walk backwards in response to a vocal order using either positive or negative reinforcement. Heart rate monitors and behavioural observations were used to assess the animals’ perception of humans on the short (just after training) and long (5 months later) terms. The results showed that the type of reinforcement had a major effect on the subsequent animals’ perception of familiar and unfamiliar humans. Negative reinforcement was rapidly associated with an increased emotional state, as revealed by heart rate measurements and behavioural observations (head movements and ears laid back position). Its use led the ponies to seek less contact with humans. On the contrary, ponies trained with positive reinforcement showed an increased interest in humans and sought contact after training. This is especially remarkable as it was reached in a maximum of 5 sessions of 1 to 3 min (i.e. 5 to 15 min) and had lasting effects (visible after 5 months). Even learning was positively influenced by positive reinforcement. Overall, horses seem capable of associating humans to particular experiences and display extended long-term memory abilities.     

Acquisition of cup drinking using previously refused foods as positive and negative reinforcement

We used previously refused foods as positive and negative reinforcement in the acquisition of cup drinking. Cup drinking increased with positive and negative reinforcement, both alone and in combination (without escape extinction), indicating that treatment of food refusal can establish some foods as appetitive stimuli whereas others remain aversive.     

Effects of response-independent negative reinforcers on negatively reinforced key pecking

Previous research has shown that presenting response-independent positive reinforcers reduces the response rate of an operant maintained by positive reinforcement. The present experiment investigated a similar effect using shock-free time as a negative reinforcer. Brief shocks were delivered in the presence of a distinctive stimulus, and pigeon's key pecks were reinforced by the occasional presentation of a 2-minute shock-free period. Extra 2-minute shock-free periods were added independently of behavior. For each of three pigeons, response rate during shock-on periods declined with added shock-free periods; the more frequently the extra shock-free periods occurred the greater the decline in response rate. This outcome is predicted by extending the Law of Effect to include negative reinforcement.     

The effect of contingency upon the appetitive conditioning of free-operant behavior

The problem of maintaining independence between response rates and reinforcement probabilities when determining the effect of varying the response-reinforcement contingency upon free-operant behavior was solved by programming local reinforcement probabilities for response and no response on a second-by-second basis. Fifty-seven rats were trained to lever-press on schedules of water reinforcement involving different values of contingency. All rats were first trained on a high positive contingency and then shifted to less positive, zero, or negative contingencies. Under these conditions, rate of lever-pressing declined appropriately when the contingency between response and reinforcement decreased or was made negative. The decline in rate produced by a zero contingency cannot be attributed to extinction, since the probability of reinforcement given the occurrence of a response was the same as for the positive contingency from which the shift to zero was made. That is, there was no change in the opportunity for response-reinforcement contiguity. It was concluded that the technique of programming local reinforcement probabilities offers promise for more critical examinations of the effects of contingency upon free-operant behavior.     

The effects of extinction,noncontingent reinforcement and differential reinforcement of other behavior as control procedures

Several techniques have been used in applied research as controls for the introduction of a reinforcement contingency, including extinction, noncontingent reinforcement (NCR), and differential reinforcement of other behavior (DRO). Little research, however, has examined the relative strengths and limitations of these "reversal" controls. We compared the effects of extinction with those of NCR and DRO in both multi-element and reversal designs, with respect to (a) rate and amount of response decrement, (b) rate of response recovery following reintroduction of reinforcement, and (c) any positive or negative side effects associated with transitions. Results indicated that extinction generally produced the most consistent and rapid reversal effects, with few observed negative side effects.     

A comparison of various forms of reinforcement with and without extinction as treatment for escape-maintained problem behavior

The present investigation compared the effects of reinforcing compliance with either positive or negative reinforcement for a participant who displayed escape-maintained problem behavior. The results indicated that positive reinforcement in the form of a highly preferred edible or leisure item produced higher levels of compliance and lower levels of problem behavior when compared to negative reinforcement in the form of escape from demands. In addition, an extinction procedure was unnecessary to achieve high levels of compliance.     

A method for the delivery of reinforcement during exercise

The purpose of this study was to develop an automated system for positively or negatively reinforcing subjects while they exercised on an electronically braked bicycle ergometer. The circuitry used to interface the subject’s pedal speed responses with the delivery of auditory reinforcement is presented, as well as a study designed to test the utility of the system. Twenty 18- to 25-year-old male subjects exercised on an electronically braked ergometer at a constant workload of 50 W for three 5-min blocks. During the first 5 min, subjects pedaled at their preferred rate, and during the next two 5-min blocks, they pedaled faster or slower than their baseline pedal speed to turn on a comedy tape for positive reinforcement, or turn off 90 dB of white noise for negative reinforcement. Circuits that interfaced the subject’s pedal response with the delivery of reinforcement are discussed in detail. Both positive and negative reinforcement increased the targeted pedaling rates, with no differences as a function of targeting increased or decreased pedal speed. This system could be used with various reinforcers, such as video and other audio equipment to increase exercise behavior.     

A review of reinforcement control procedures

A demonstration of the effects of reinforcement requires comparison of response rates in the presence of a contingency with those in another condition that controls for the influence of extraneous variables. We describe several control conditions that have been used in evaluating the effects of positive and negative reinforcement. The methodological rigor and practical utility of each type of control condition are discussed, and recommendations for the use of these conditions are presented.