The Differences Between Classic And Operant Conditioning

Research into learning often concentrates on what is being learned, but the emphasis of learning theory is on the process of learning itself, in broad terms ‘how is it learned’. When we focus on the end product of learning – ‘what’ is being learned – we might assume that the learning itself is deliberate and perhaps directed by a teacher. This is not always the case as learning can take place through the observation of others, but also when no other being is involved, e.g. if two events happen regularly in the same environment, we can learn they are related without being conventionally taught. The issue of learning is one of the most researched and discussed areas in psychology and the behaviourist approach has exerted a major influence in this area of research. (Gross, 1987). Associated with this research is the observation of two types of behaviour – respondent, which is an automatic behaviour triggered by particular stimuli, and operant behaviour, which is a voluntary behaviour not tied to any stimuli. Furthermore, related to these behaviour types is the distinction between classical and operant conditioning, which display important differences, although they are both Stimulus-Response (S-R) approaches to learning (Gross, 1987).

During the 1890s, Russian physiologist Ivan Pavlov was researching salivation in dogs when exposed to food as a stimulus. To facilitate his research, Pavlov developed a technique for collecting the saliva of dogs so secretions could be measured. However, during his research, Pavlov noticed that the dogs he observed would begin to salivate prior to food being given, including when they could see food or the feeding bucket or even on the approach of the lab assistant who fed them. Moving on from these observations, Pavlov noted that when paired with a stimulus that normally produces a particular response (an unconditioned stimulus or UCS), a stimulus that would not normally promote this response (a conditioned stimulus or CS) will eventually do so.

To demonstrate this theory, Pavlov used his previous observation of the interaction between food and salivation and then linked this to the sound of a ringing bell whenever food was presented. The taste or smell of food automatically creates salivation in a dog, but usually, the sound of a bell will not promote this reaction. A dog does not need to learn to salivate when it sees food; this behaviour is an automatic, reflexive, biological response, that happens naturally and is not learned. The food is therefore referred to as the Unconditioned Stimulus (UCS) and the salivation is an Unconditioned Response (UCR). However, when the sound of a bell is paired with the presentation of food a sufficient number of times, a dog will start to salivate at the sound of a bell before food is presented. The sound of a bell ringing does not naturally cause salivation, it is, therefore, referred to as a Conditioned Stimulus (CS) and the salivation caused by hearing the bell ringing is known as a Conditioned Response (CR). Figure 1 sets out the basic procedure of classical conditioning. Other types of conditioned stimuli (CS) can be used including lights, buzzers, etc. In the case of dog training, a clicker is often used – with or without food – to facilitate conditioning in dogs (Chiandetti et al., 2016; Feng et al., 2018).

Four different types of conditioning have been identified by varying the relationship between the CS and the UCS. These types of conditioning are Delayed (or Forward) Conditioning, Backward, Simultaneous and Trace. Not all of these types of conditioning are effective when used to train animals. Delayed Conditioning is the type of conditioning habitually used to train animals in the laboratory and elsewhere. During this conditioning, the CS is presented approximately half a second prior to the UCS and remains ‘on’ or present until the UCR appears. The conditioning is said to have been successful when the CR appears before the UCS is presented. Backward Conditioning is generally used in advertising aimed at humans, as it uses the process of presenting the CS after the UCS which has little effect on animals (Bouton and Bolles, 1980). Trace Conditioning involves the presentation and removal of the CS before the UCS is presented. As this causes only a ‘memory trace’ of CS, in some cases, Trace Conditioning is a weaker form of conditioning and less resistant to extinction, when compared to Delayed or Simultaneous Conditioning (Church et al., 1956; Ellison, 1964). Simultaneous Conditioning involves the CS and UCS being presented at the same time and for the same period of time. Conditioning is said to have happened when the CS independently produces the CR (Gross, 1987).

Classical (as well as operant) conditioning are complex processes that can be influenced by other phenomena. These phenomena are termed generalisation and discrimination. During generalisation, there is a spontaneous transfer of the CR to stimuli that are similar to, but not the same as the original CS. In the experiment Pavlov used, this entailed presenting the dogs with bells of different pitches. The dogs are usually salivated when presented with bells of higher and lower pitches to the pitch of the original CS bell, but food was only presented when the original bell is rung. Eventually, however, when the dogs were continually presented with bells of different pitches from the original, the CR weakened and gradually stopped altogether. In this way, the dogs were said to be exhibiting discrimination, specifically spontaneous discrimination. Pavlov also discovered that as part of the original conditioning exercise, dogs could be trained to salivate to a specific bell pitch if food was presented when a high-pitched bell rang and withheld when a low-pitched bell rang. Pavlov called this discrimination training (Gross, 1999, 1987). If conditioned dogs were repeatedly exposed to a bell ringing without presentation of any food, the salivation CR grew weaker, appeared to be ‘unlearnt’ and was eventually considered to be extinguished. This was shown not to be the case as when the dogs were removed from laboratory conditions for an hour or so and then returned and presented with a bell, they again began to salivate even when food was not presented. The return of the CR for salivation was termed spontaneous recovery suggesting that extinction does not totally erase the original learning/conditioning, rather when the CS is continually presented without the UCS, the CS is inhibited or supressed.

Sadly, as Pavlov pushed further the research concerned with discrimination and generalisation, his experiments began to have serious, deleterious effects on the dogs involved. After training dogs to salivate to the shape of a circle, but not to an ellipse and then gradually changing the ellipse until it almost appeared to be a circle, the dogs involved, not knowing how to respond, started to display what Pavlov called experimental neurosis, which included refusing to eat, whining, trembling, defecating and urinating (Gantt and Gantt, 2007; Gross, 1987; Windholz, 1991).

In contrast, B. F. Skinner’s atheoretical approach to behavioural research was aimed at non-selective and unprejudiced observation, rather than trying to explain it through theory. His work concentrated on the suggestion that almost all behaviours are determined and controlled by the consequences of positive and negative reinforcements; with positive reinforcement being rewards and negative reinforcements those consequences that we work to avoid or escape from. Skinner’s work with rats and pigeons showed that positive reinforcement (reward) was more effective in controlling behaviour than negative reinforcement or punishment. This work was a modification of Thorndike’s ‘Law of Effect’ theory, which suggested that behaviour that was followed by a pleasant consequence, was likely to be repeated and behaviour followed by an unpleasant consequence, was likely to be avoided in future (McLeod, 2018). Skinner advocated positive reinforcement as a way to attain socially desirable goals, proposing that behaviours are responses that have been reinforced in the past and are determined through the pursuit of things that have been positively reinforced. To demonstrate this Skinner used a form of puzzle box (the ‘Skinner box’) which in the case of rats, offered a lever that when pressed delivered a food pellet to a tray below the lever.

Gross (1987) suggests that Skinner’s analysis of behaviour “…requires an accurate but neutral representation of the relationship (or contingencies) between: (a) antecedents (the stimulus conditions, e.g. the lever, the click of the food dispenser, a light that may go on when the lever is pressed); (b) behaviours (or operants, e.g. pressing the lever); and (c) consequences (what happens as a result of the operant behaviour).” The consequences of behaviour can be positive reinforcement, negative reinforcement or punishment. Whilst they work in different ways, both positive and negative reinforcement can have a strengthening effect on behaviour, i.e. making it more likely to happen. During positive reinforcement, the animal being observed is presented with something it likes (e.g. food or a toy). Conversely, negative reinforcement involves the removal or avoidance of a painful situation – an aversive state (e.g. the loosening of a prong collar when a dog stops pulling on a lead) (Madson, 2019). Punishment involves the presentation of an aversive stimulus, but has the opposite effect to reinforcement techniques and weakens behaviour. Although negative reinforcement and punishment sound similar, they differ in that during reinforcement an aversive stimulus is removed or avoided, but during punishment, the aversive stimulus is commenced (e.g. a shock from an electric collar when a dog is observed chasing livestock). From his research Skinner concluded that positive reinforcement (and negative reinforcement) had far more influence on behaviour than punishment because punishment cannot teach anything new; it only makes certain behaviours less likely (Cooper et al., 2014).

In conclusion, the essential difference between classical and operant conditioning is the distinction between respondent and operant behaviours. Whilst Pavlov claimed that behaviour is elicited by specific stimuli, Skinner argued that behaviour (animal and human) is a consequence of how animals operate in their environments by learning the consequences of their actions. In Skinner’s view, the learner was a far more active participant than the automatic, involuntary respondent to stimuli (unconditioned before learning and conditioned after) suggested by Pavlov’s work. There is a certain inevitability to classical conditioning; with the animal given little choice in how to respond. Operant conditioning focuses on strengthening or weakening behaviours through responses that are voluntary and less predictable, as behaviour is emitted by the animal and not elicited by a stimulus, and the response is a function of the consequences of past behaviours (Cherry and Susman, 2020).

It should be noted that given current research ethics guidelines neither Pavlov’s nor Skinner’s research would likely be given the go ahead today.