Theories Of Learning: Behaviourism And Multiple Intelligences

Behaviourism

The Theory

The basics

History:

Scientific experimental research on humans started in the nineteenth century when Wundt set up a psychological laboratory in Germany in 1879 (Child, 2010). At the turn of the century, Pavlov, Watson and Thorndike started studying how animals and humans behaved in controlled circumstances (Child, 2010).

Watson was the first to conclude that the temporal proximity of successful stimulus-response (S-R) connections was proportionate to their likelihood of reoccurring (Child, 2010). If a person (or animal) attempts to solve a problem, the next time they face a similar problem, they are likely to repeat their prior actions that came shortly before successfully solving it. The strongest S-R bonds are formed when stimulus and response occur simultaneously.

While accepting Watson's theory, Thorndike thought that rewarding responses were more important than their timing, and S-R bonds are generated when an action is reinforced by a satisfying result (Child, 2010). Therefore, while showing a child a picture of an apple whenever they say the word may help them make the association, giving a child an apple whenever they say the word would be more effective.

Pavlov also thought of behaviour as responses caused by stimuli but limited his research to reflex actions - famously the salivation reflex in dogs (Child, 2010).

Many early behaviourists thought S-R bonds were formed only with elicited responses and not emitted responses (classical conditioning). However, Skinner concentrated on creating S-R bonds by rewarding or punishing emitted responses (operant conditioning) (Child, 2010).

Because Skinner is the best-known behaviourist of our time, and his theories are the most influential in classrooms today, we will focus hereafter on his specific theory of Behaviourism.

Rationale:

Skinner's theory started with the question, 'How can we anticipate and affect people's actions?' He viewed the chain of S-R as:

External stimulus ➡ Internal feelings ➡ External response

Therefore, if we want to anticipate or affect a person's actions, we must either appropriately stimulate them or observe their internal feelings.

Mentalists attributed what a person does to internal feelings. However, Skinner baulked at this idea

So Skinner changed his S-R chain to:

External stimulus ➡ External response

By this, he was not denying the existence of internal feelings but viewed them as a black box that could be ignored for functional analysis. Psychologists could then study behaviour scientifically with controllable inputs and measurable outputs.

To summarise Skinner's way of thinking, "If we know that a child has not eaten for a long time, and if we know that he therefore feels hungry and that because he feels hungry he then eats, then we know that if he has not eaten for a long time, he will eat" (Skinner, 1974, p13).

Method:

As mentioned above, Skinner favoured the approach of Operant Conditioning over Classical Conditioning. Classical Conditioning dealt with creating S-R bonds with elicited responses (making a dog salivate by presenting it with food). However, Skinner moved beyond reflex actions to S-R bonds with emitted responses (a bird happens to walk in a circle and learns this results in a reward). This is why Skinner's conditioning is called Operant Conditioning, as the subject (the operant) has some degree of freedom in their responses.

He then used positive and negative reinforcers to produce S-R bonds - a particular action would lead to a reward or the cessation of a negative stimulus such as captivity, leading the subject to repeat that action in future.

A Black Box

In the novel 'Matter', when the Regent asks his Field Marshal if he trusts the Oct (an alien race), he replies,

Much of Skinner's thinking relies on viewing all internal thoughts as 'alien'. Either someone will react in a certain way to a stimulus, or they will not, and to speculate as to what internal processes guide their actions is a fool's errand. This thinking all hinges on our internal thoughts being a black box, as if they are known, then surely mentalism provides better answers.

In interviews with tennis players, they do very poorly at describing how they successfully hit a ball. Slow-motion footage reveals physical movements the players are not consciously aware of (Gladwell, 2006). It would seem our internal thoughts are sometimes a black box, even to ourselves.

Nevertheless, whether these internal thoughts are known or not, it would appear they are still relevant and can not be simply stepped over as Skinner suggests. Pritchard notes that:

Thus, it appears Skinner's theory may have more usefulness in the study of reflexes than in application in the classroom setting.

Speculative and Circular

Skinner often makes statements about internal feelings without argument or evidence. For example, he claims without evidence that feelings are the result of only genetic and environmental histories; feelings are not what cause reinforcement but merely accompany it; and that feelings do not cause behaviour but have a common cause (Skinner, 1974). This makes his theory appear speculative and circular.

Skinner admitted himself that "Much of the argument goes beyond the established facts" (1974, p19), and Sotto points to the circularity that "a reinforcer is defined as that which causes a behaviour to be repeated" (2007, p38).

Skinner defends himself by saying that if speculation is precluded from science, then much of astronomy and atomic physics would be cast out.

What Skinner seems to have failed to recognise is that while speculation may be used in hypotheses, theories should be founded on experimental evidence and sound reasoning.

Materialistic

Skinner saw feelings as the result of genetic and environmental histories (1974). He seems to share common ground with the techno-prophet Ray Kurzweil, who said:

These lines of thinking have their roots in materialism. If nothing exists except the material things controlled by the rigid laws of physics, humans and machines are comparable, and feelings can be calculated from genetic and environmental histories.

John Searle, a professor of mind at the University of California at Berkeley, thinks otherwise.

While Skinner may think shuffling symbols is all humans do, the existence of something more, consciousness, has been known for centuries:

Even if Skinner's materialistic position is to be accepted, we are then faced with the logical conclusion that we cannot know if Skinner's theory is correct. The evolutionist Haldane put it very well;

Humans and Animals

Aside from applying conclusions from experiments with animals to humans, Skinner also ascribed very animalistic intelligence to us. For example, he said:

It is easy to think of circumstances where the manner of looking for a match has not been previously reinforced. For example, when you open a matchbox and find it empty, you do not close the matchbox and then open it again in an attempt to find a match inside. Humans use a depth of reasoning not explained by conditioning.

Thorndike, the father of Skinner's theory, said that he began his work "to give the coup de grace to the despised theory that animals reason" (in Hilgard & Bower, 1975); he did not think his experimental results applied to humans.

The incongruence of the observed behaviour of animals and humans should be well understood in any application of Skinner's theory.

Free-Will

Skinner's theory seems to discount the concept of free will. He states unequivocally that "if he has not eaten for a long time, he will eat" (1974, p13), leaving no room for choice. But what if the individual chooses not to eat? What of those around the world on hunger strikes?

There is a much simpler explanation for Skinner's experimental observations - subjects do not respond because of their S-R bonds acquired through conditioning, but instead, they respond because they remember the previous positive and negative re-enforcements. When you go to the cupboard to find food, it is not because you were conditioned to but because you remember where you put the food, and you remember that eating food removes your hunger. This remembering leads to your choice to respond - or your choice to starve yourself.

An experiment in verbal positive reinforcement showed that those with conscious knowledge of the desired behaviour performed vastly better than those who were not conscious of the goal (Hilgard & Bower, 1975). This indicates that the subjects chose to respond in the way they did.

However, another experiment which rewarded minute muscle responses showed a positive correlation (Hilgard & Bower, 1975). This indicates the possibility that unconscious physical actions may be conditioned (though see the discussion on Thin-Slicing later).

If proof were needed of the possibility to override severe conditioning with free will, look towards the Vietnam war:

Modelling and Thin-Slicing

As we have already seen, Skinner viewed behaviour as:

External stimulus ➡ Internal feelings ➡ External response

before simplifying this to:

External stimulus ➡ External response

This simplification only works with the assumption that you can never have behaviour created by spontaneous internal feelings.

Modelling:

One human experience which seems to work against Skinner's assumption is the internal modelling we do which often changes behaviour.

For example, a person scared of spiders may visit a therapist who gets them to imagine (to model) safe, even pleasurable, experiences with spiders. Their behaviour of acting fearfully can thus be modified without external stimulus (though behaviourists may simply move the goalposts and claim the therapist was the external stimulus). Or perhaps a person is at a train station saying farewell to a loved one - they cry not because the person has already left but because they model in their heads what it will be like after they have left.

Experiments using subconscious suggestion have shown that simply thinking about words we associate with being old makes people walk more slowly, without conditioning. The same is true of 'aggressive', 'polite', 'clever' and 'stupid' word associations and their behavioural outcomes (Gladwell, 2006). This points towards internal, subconscious modelling of reality affecting human behaviour. The S-R approach would only expect these results with prior conditioning and re-enforcing behavioural outcomes - both of which were not present. To remove the argument that the stimulus is external, not internal - simply choosing to think clever thoughts before an exam can improve your results without exposure to the words (Gladwell, 2006).

People's different internal models go some way to explain the behaviour of psychiatric patients - something that Behaviourism fails to explain adequately.

Thin-Slicing:

A related human experience to modelling is Thin Slicing - "the ability of our unconscious to find patterns in situations and behaviour based on very narrow slices of experience" (Gladwell, 2006, p23). This theory does not refute Skinner's view of conditioning but provides an alternative that incorporates the phenomenon of modelling.

A simple game was played in an experiment at the University of Iowa. Players were presented with four face-down decks of cards - two blue and two red. Each card the player turns over either loses or gains them money. What players did not know is that blue cards were better on average than red cards (Gladwell, 2006).

After fifty cards, players had a 'hunch' that blue cards were better than red, and after eighty cards, they could articulate why blue cards were better than red cards. However, after only ten cards, players started to have a sweat response to the red cards and unconsciously started turning over more blue cards.

These findings can be explained through Behaviourism. However, the alternative view of Thin Slicing - that players subconsciously modelled the situation and internally changed their behaviour - can equally explain the findings and why phobias can be "thought away" and people can miss their lovers while they are still there.

Higher Order Skills

When behaviourists talk about learning, they usually mean "the ability to do something that one could not previously do. It follows that 'learning' results in 'a change'" (Sotto, 2007, p34). So when a behaviourist wants to measure learning, they are looking to measure a changed behaviour that leads to success.

Contrary to this viewpoint is Lashley's findings (in Sotto, 2007), who trained a rat to obtain food by performing specific movements. He then removed parts of the rat's brain, preventing it from carrying out those movements, after which the rat obtained food by carrying out different movements. Thus, the rat had not learned a set of behaviours that commonly had the consequence of food but rather had learned the model of how to obtain food and invented behaviour that fitted the model.

An experiment by Tolman (in Sotto, 2007) similarly showed that rats in a maze do not learn the behaviour of how to escape, but they learn to model the maze and then invent behaviour to reach the exit.

This is why Programmed Manuals/Teaching Machines were a failure: "…except where routine skills are to be learnt, they have now all been discarded" (Sotto, 2007, pp35-36). Behaviouristic teaching does not lead to real learning — it leads to repetition. 'Do the right thing for a reward and keep repeating the same action'. The higher-order skills of application, analysis, synthesis and evaluation are beyond the scope of behaviourism.

Classroom Implications

Despite the criticisms above, Behaviourism has seemed to be applicable in classrooms over the decades. We will now attempt to find useful applications with the proviso that it is limited to simple behavioural modification and not deep learnings.

Types of Learning

Since Behaviourism can not teach higher-order skills, it should be used in conjunction with other learning theories.

A teacher's job is not only to teach understanding but to teach behaviour. If they are to change the observable behaviour of a pupil who shouts out or refuses to work, then Behaviourism is a valuable tool. However, if they are to develop a pupil's understanding, Behaviourism would be the wrong tool.

Reward vs Punishment

Before comparing rewards and punishments, we must first clarify the difference between punishment and negative reinforcement. Whereas punishment is designed to remove existing behaviour, negative reinforcement is designed to elicit behaviour - such as one that can be caused through pain or "the punishment for not behaving" in the desired way (Skinner, 1974, p62).

Punishment, though designed to remove behaviour, is not the reverse of conditioning (be the conditioning negative or positive) as its effects also create new avoidance behaviours. Instead of the subject ceasing the behaviour, they simply try to continue with the behaviour while avoiding the punishment (Skinner, 1974). Punishing a cat for jumping on the kitchen work surface teaches them only to do it when their owner is out of the house.

Even early behaviourists such as Thorndike realised that the positive effects of rewards were much stronger than the negative effects of punishment.

Knowing this, teachers should always look for ways to positively reinforce pupil behaviour rather than concentrating on punishment - something that has seen a swing in schools from the cane to house points systems. Sadly most school discipline policies are still much longer than school reward policies.

One objection to using rewards over punishment in operant conditioning is that the subject must first be active in a way that produces a reward, but "…in a conventional classroom, the learners will probably be sitting on their backsides and listening… there would be very little in their behaviour that could be reinforced" (Sotto, 2007, p35). This is often the desired behaviour though, and thus should be reinforced. But what if the desired behaviour is not present?

This is why Skinner concluded that each step of learning should be short and follow from prior learning, and others conclude rewards may be used to reinforce approximate behaviour at first before only rewarding the specifically desired behaviour later (Child, 2010).