Analysis Edward Lee Thorndike's Behaviour Experiments

Prior to Edward Lee Thorndike’s landmark “experimental analysis of behaviour” in 1898, the study of the psychology of learning lacked a clear, defined research methodology. Research findings were very much subjective, lacking the quantitative evidence that would give them scientific credence and avoiding impreciseness and ambiguities in their interpretation. People were asked to “look inside their minds and describe what they were thinking.” (Ormrod, 2008). This method was called introspection, which by its very nature, is devoid of objectiveness.

Thus emerged Thorndike’s groundbreaking “puzzle box” experiments with cats, dogs, and chicks, culminating in his doctoral dissertation on animal intelligence in 1898. Through his experiments, the study of learning took on a more objective approach, where the emphasis was on the observables rather than the non-observables. The basis of his experiments was that “behaviour” is an observable phenomenon and thus is measurable. It led to the birth of the behaviourist movement where research looked primarily at behaviour rather than cognition, given that mental processes (like insight or introspection) cannot be objectively measured with any level of reliability.

Thorndike’s Experiments

To circumvent the subjective nature of research then, Thorndike developed some experiments on the “learning” phenomenon whose results were beyond subjective interpretations. The idea behind them was to observe the behaviour of a hungry animal trapped in puzzle boxes in its attempt to escape. He conducted the experiments using three animals, viz. cats, dogs, and chicks. Food, functioning as the “temptation” factor (stimulus), was placed outside the box in full view of the animal.

He started with a cat. For it to escape, the cat had to manipulate a device that would open the door in order to get at the food. There were altogether 15 of these boxes, each with different escape mechanisms, ranging from a wire loop, lever, and a treadle which had to be depressed for it to escape. The variety of mechanisms was (presumably) to observe any consistency in behavioural mode despite the variables presented by the escape devices.

In its attempts to escape, Thorndike closely observed the cat initiating numerous, apparently random movements, seemingly driven by impulse rather than reasoning. Through the cat’s struggles, it would eventually trigger the release mechanism. This was achieved more by chance through “trial and error” rather than by any reasoned action of the cat.

The cat was repeatedly put into the box. Thorndike observed that what was initially random and chaotic behaviour became relatively more orderly and efficient. In other words, he noticed gross changes in behaviour when the same set situation was repeated. In addition, he recorded the time it took for the cat to escape from its confinement. He then plotted a “time-curve” graph which enabled him to not only analyse the animal’s speed of learning, but also the rate of learning.

Thorndike repeated the same puzzle box experiment with dogs and chicks, with minor practical adjustments to the procedure. Again, time-curves were plotted and compared to those exhibited by the cats.

However, in some experiments, Thorndike varied the escape mode. Instead of the animal physically manipulating a release device, he would, for example, remove a chick from a box whenever it preened its feathers; or he would open the door when a cat licked or scratched itself. He observed that the chick started preening itself when it was re-placed into the box, indicating the animal’s instinctive desire to be let out of the box. Time curves were plotted and compared to the curves obtained by the escape device mode.

There were also certain other variations in the way the experiments were conducted. One such variation was placing the same cat that was used in a particular box into another differently designed box. He then observed how the cat reacted to the changed environment. In another variation, Thorndike allowed the animal to observe another member of its species in its attempt to escape. He wanted to find out if animals could learn by imitation.

Summary of Main Findings

Overview

The overarching finding of Thorndike’s experiments was that there are responses (explicit behavioural changes) when there is a stimulus (food). In that sense, there is a stimulus-response (S-R) connection in how animals “learn”. How well the animals learnt is determined by the experience it gained from the same experiment conducted repeatedly. In effect, this perspective of learning (connectionism) emphasized the role of experience in the strengthening and weakening of the S-R connections. In general, it could be said that the essence of intellectual development depends on how strong this S-R connection is.

The main findings from Thorndike’s experiments in animal intelligence would form the basis for his formulation of theories related to learning. They led to further research into the more complex intellectual abilities of humans, eventually leading to the development of modern era comparative psychology.

Findings from the Experiments

The main findings from Thorndike’s puzzle box experiments can be summarized below:

Gross changes in behaviour

When presented with a stimulus (S), the animal displayed an overt response (R), i.e. food (the stimulus) elicited a reaction in a hungry animal. In other words, Thorndike noted explicitly observable behavioural changes when an S-R setting is presented to the animal. The “gross changes” described by Thorndike manifested itself by the animal’s impulsive actions in what he described as trial-and-error learning in order to escape from the box.

Learning is gradual

Time curves were plotted in all of Thorndike’s experiments involving cats, dogs and chicks. The curves were a measurement of the time required for the animal to escape over repeated number of trials ranging from 24 to 117. The graphs allowed Thorndike to not only obtain the escape times, but also the rate of learning, which was represented by the slope of the curves.

The time curves presented by all the animals showed remarkably similar patterns of behaviour. Analyses of the curves proved that learning took place gradually, i.e. in incremental steps rather than huge jumps. This was seen in the gradual reduction of escape times over the total number of trials. In one of the experiments, for instance, escape times for a cat varied from a high of 160 seconds to a low of 6 seconds over 24 trials, indicating that it had been steadily more efficient in its attempts to escape. Thorndike alluded to this when he described its initially random, chaotic demeanour gradually becoming more orderly and deliberate. It also proved that in some innate way, the animal had “learnt” from past experience.

Effective behaviour diminishes ineffective behaviour

By analysing the time curves, Thorndike concluded that effective behaviour, i.e. actions that could lead to a satisfactory end, persisted. Conversely, ineffective behaviour diminished. In effect, there was a “stamping in” of some behaviours and a ‘stamping out” of others (Ormrod, 2008). This phenomenon was demonstrated in the way the animals became more efficiently adept in their escape attempts as proven by the quicker escape times. For example, in the experiment with a chick, the initial escape attempt was arbitrary and disorderly. But after many repetitions, the chick “learnt” to escape quickly with little of the initial chaos. As Chance (1999) puts it: “Eventually … the bird performed the act required for escape as soon as it was placed in the enclosure.” In short, ineffective behaviour petered out. It could also be said that generally, other than through experiential learning, positive learning generates further positive learning.

Generalization of behaviour

Thorndike also found that an animal that had learnt to escape via a particular means, e.g. by clawing, tended to employ the same means when placed in a different box with a different escape device. It could be inferred that animals generalize when first put in circumstances different than what they are familiar with, with their immediate prior experience influencing how they subsequently behave. It is only after they have adapted to the new environment that they adopt a new learning approach.

Discriminatory behaviour

Another aspect of learning that Thorndike noted from his experiments was that an animal is capable of being discriminatory. This was evident, in one of his experiments, when (i) he made a statement to feed it and indeed fed it, and (ii) he made a statement to feed it, but he did NOT feed it. There far fewer errors in situation (i) than there were in (ii). This illustrated yet another perspective about the learning process that Thorndike delved into.

Learning by observation

Thorndike asserted that animals, at least other than primates, do not learn by imitation, even after observing the successful behaviour of other members of their species. He effectively concluded that imitation could be discounted as an aspect of an animal’s successful behaviour.

Theories propounded by Thorndike

Thorndike’s experiments on animal intelligence resulted in the formation of a body of theories related to the learning process and laid the scientific foundation for educational psychology. These learning theories were bound together by the theory of connectionism. The principal credo of Thorndike’s connectionism is that “learning is the result of associations forming between stimuli (S) and responses (R)” (http://tip.psychology.org/thorn.html).

Connectionism debunked the dominant view held by psychologists before Thorndike that animals learnt by the association of ideas, i.e. they could logically relate events and reason out solutions to a problem. Rather, through his experiments, Thorndike established that animals are equipped with “action impulses”. It was these impulses, responding to the stimulus (S), that activated the trial-and-error behaviour of the animals trapped in the puzzle boxes. Their eventual successful escape was the result of the gradual learning they experienced arising from actually performing the act. It derived from an innate understanding that the correct actions (affirmative response) produce the desired effect (getting to the food). In short, Thorndike believed that learning requires no ideas in the head of the animal; only the performance of the act itself was essential for a satisfactory consequence.

The learning theories propounded by Thorndike, being subsets of connectionism, can be summarised as follows:

The Law of Effect

“Responses to a situation that are followed by satisfaction are strengthened; responses that are followed by discomfort are weakened” (Ormrod, 2008).

This theory basically emphasizes the role of responsive actions and their consequences. In the act of responding, the impulse that produced the successful act would be “stamped in” (strengthened), whilst all other non-successful impulses would be “stamped out” (weakened). Eventually the “stamping in” would become habitual responses to that same situation.

In the puzzle box experiments, the S-R connection was established because the response resulted in a satisfying consequence (escape from box). This response was strengthened, as seen when the animal reduced its vain actions (e.g. clawing and scurrying aimlessly) in the repeated experiments, as shown by the much faster escape times.

Applied to human learning, this law implies that positive stimuli breed correspondingly positive responses. In practical terms, in schools or any learning institution, students should be provided with an environment that is conducive to learning. Libraries, for example, in general provide this conduciveness. As Ormrod (2008) puts it: “Students should experience academic tasks in contexts that elicit pleasant emotions … rather than in contexts that elicit anxiety, disappointment, or anger.” This statement is consistent with the premise of Thorndike’s Law of Effect that satisfactory consequences strengthen the response.

The Law of Readiness

“A series of responses can be chained together to satisfy some goal which will result in annoyance if blocked.” (http://tip.psychology.org/thorn.html)

This theory arose in an experiment where a chick had to execute a sequence of actions to trigger a series of release devices for it to escape confinement. This sequence of acts is today known as “a response chain”.

Extending the aspect of “satisfying some goal” to humans, it could be surmised that optimal learning takes place only when someone is ready to act; in so doing, the consequences are satisfying. On the other hand, forcing someone to act when he is not ready will be annoying. “Forcing” can be interpreted as interfering in someone’s goal-directed behaviour. Interference that leads to unwilling behaviour causes frustration.

An analogy could be made of a child who is force-fed to eat vegetables. He may grow up to hate eating vegetables if it is forced upon him. However, if he himself is ready to eat them, the act will more likely lead to satisfaction. In the long run, vegetables are a must in his meals; not having them may lead to annoyance.

On a broader perspective, the level of readiness could be linked to a major factor in the efficacy of learning, viz. motivation. At its basic level, motivation is “some kind of internal drive which pushes someone to do things in order to achieve something.” (Harmer, 2007). When a person is sufficiently motivated to learn, the aforesaid “internal drive” should ready him/her to learn. In the context of the classroom, a conducive learning environment (S) can also motivate (R) students (the S-R connection in the Law of Effect). Hence it can be said that the stimulus (S) breeds a response (R) (of readiness to learn); in turn the response (R) breeds another positive response (motivation). Here, there is a chain of positive responses in the learning process.

The Law of Exercise

“Stimulus-response associations are strengthened through repetition.” (Wikipedia, 2009)

There are two sub-laws arising from the theory:-

• The Law of Use: “The S-R connections are strengthened as they are used.”
• The Law of Disuse: “The S-R connections are weakened as they are not used.”

In effect, this theory stresses the importance of repetition in the learning process. This was how the animals strengthened the S-R response progressively in Thorndike’s experiments, resulting in the faster escape times evident in the time-curves. The “exercise” in the law refers to practice, as in the commonly-held adage “Practice makes perfect”. This is especially true, not only cognitively, but also in skills where psychomotor and kinaesthetic abilities are critical to the learning. For instance, the more successful golfers are more likely to be the ones who hone their skills through longer periods of repetitive drills, all other factors being equal. Motor mechanics achieve a higher level of competency through constant and repetitive work.

The importance of repetitive practice in learning cannot be overemphasized. A student can achieve a greater mathematical proficiency through tireless repetitive practice. In language learning, repetition has always played a part … to provoke the structuring and re-structuring of ‘noticed’ language (Harmer, 2007). Hence teachers in language classrooms put their students through choral drilling for them to internalise not only grammatical structures, but also the sentence intonation.

The three laws of learning above are interestingly linked to each other. Connections are strengthened because S-R pairings occur many times (Law of exercise) and then rewarded (Law of effect) as well as forming a single sequence of actions (Law of readiness). Re-stating the link in another way, it can be said that in learning, motivation (readiness) is driven by incentives (effect). Both motivation and incentives can drive the learner towards practice (exercise) until ultimate success is achieved.

Thorndike’s perspective of animal learning

The MacMillan English Dictionary (2002) defines “reasoning” as: “the process of thinking about something in an intelligent and sensible way in order to make a decision”. The Concise Oxford English (2002) dictionary defines “thinking” as: “using thought or rational judgement; intelligent”. But are both these definitions applicable universally, i.e. applicable to humans and animals?

In a general sense, it may appear so. After all a process of ‘thinking’, in whatever way, has to take place before an animal acts. But does an animal “think” by using rational judgement? On the premise that animals do not rationalise, it follows then that animals do not “think”, at least not in the way that we humans do. Barrow and Woods (2006) declared that rationality is inextricably tied up with the notion of thinking. So it begs the question: How do animals ‘think’? Given that thinking is needed for learning, how do animals learn?

Thorndike, in his doctoral dissertation, asserted that “animal learning has nothing to do with reasoning or the association of ideas”. Rather, he declared that “it (learning) occurs as a result of ‘trial and accidental success’.” He concluded this based on his puzzle box experiments where success (of escape) was due to the animal’s trial-and-error actions rather than a reasoned, logical, i.e. “thinking”, approach to the problem. According to Thorndike, the outwardly random, chaotic behaviour of the animal was due to the activation of its innate “action impulses” responding to the stimulus. There was no evident “association of ideas” in the solution. In layman’s terms, animals do not “put two and two together” (as humans do) in a rational, logical way when they act, i.e. there is no reasoning. If animals could reason, then learning should be abrupt, NOT gradual as proven in Thorndike’s time curves.

What Thorndike offered, on the other hand, was that instead of the “association of ideas”, animals learnt through the “association of sensations”. This perspective was diametrically opposed to the views subscribed by comparative psychologists before him. They had contended that animals had cognitive insight, they being able to establish ideas and employ reasoning in their actions. Thorndike’s experiments effectively debunked that long held contention. He said that “there was no solid evidence that animals grasped ideas or learned through reasoning” (Chance, 1999).

In summary, Thorndike put forth four arguments in support of his beliefs. These arguments were empirically backed by the main findings of his puzzle box experiments. Chance (1999) expressed them thus:

1. The behaviour of animals … is impulsive and apparently random, not systematic and logical. As enunciated above (and in other sections), when presented with a stimulus, the response of a trapped animal is chaotic; it follows that there is no contemplation or thoughtfulness, therefore no insight. “Impulse” rather than “thought” drives its actions.
2. The change in an animal’s behaviour is gradual, not abrupt. This assertion was validated by the time curves. The curves established that animals learn in incremental steps, not in sudden surges. This implies that there is an “absence of reasoning” (Chance, 1999).
3. The animals show no sign of understanding between action and consequence even after they have learnt to escape from the box. This arose from an experiment when a cat, which had escaped by pulling a loop, would repeat the same action, even when the loop was absent in the repeated trials. It shows that the cat’s response had been strengthened due to earlier successes (The Law of effect). It again confirms Thorndike’s view that animal learning does not involve reasoning.
4. Animals learnt only if they performed the necessary act themselves. As described in 3.2.6 – Learning by observation, animals do not learn by observing and imitating models. Neither would they learn if they are prompted to perform the act, as Thorndike discovered when he repeatedly assisted the cat to pull the release device. When left on its own, the cat could not associate the loop as a means of escape. This indicates that, unless the animal acted on its own volition, learning would not take place (the Law of readiness). It stands that animals cannot reason their way through a problem.

Conclusion

It has been over 100 years now since Thorndike’s pioneering work in the study and psychology of learning, resulting in a learning perspective referred to as connectionism, emphasizing on behaviour as opposed to cognition. It laid the foundation for latter day psychologists to conduct a more objective, scientific research into the process of human and animal learning.

Thorndike’s puzzle box experiments primarily focussed on the S-R relationship that provided the catalytic impetus to learning. It led to the development of Thorndike’s theories, which expounds how learning takes place. Through his experiments, Thorndike also disproved the views held by other psychologists that animals possess “insight” in their learning process. All in all, Thorndike’s research provided a fresh perspective of the learning process which led to further research by others. In a way, this scenario is an instance of the S-R environment.