Connectionism

Edward L. Thorndike (1874–1949) was a most eminent psychologist of the United States of America, whose theory of learning—termed connectionism—held considerable sway within the United States during the first moiety of the twentieth century (Mayer, 2003). In contradistinction to many early psychologists, his interests lay in the realms of education, most especially learning, transference, individual disparities, and intelligence (Hilgard, 1996; McKeachie, 1990). He adopted an experimental methodology in the measurement of pupils' attainment outcomes. The extent of his influence upon education is mirrored in the Thorndike Award, which stands as the highest honour bestowed by the Division of Educational Psychology of the American Psychological Association for distinguished contributions to the discipline of educational psychology.

Thorndike's magnum opus doth comprise a trilogy entitled Educational Psychology (Thorndike, 1913a, 1913b, 1914). He did posit that the most fundamental form of instruction doth entail the formation of associations (connections) 'twixt sensory experiences (perceptions of stimuli or events) and neural impulses (responses) that manifest themselves behaviourally. He maintained that instruction oft occurs via trial and error (selecting and connecting).

Thorndike commenced his study of instruction with a series of experiments upon animals (Thorndike, 1911). Animals in problematic circumstances endeavour to attain a goal (e.g., procure sustenance, reach a destination). From amongst the myriad responses they art capable of, they select one, execute it, and experience the consequences thereof. The more oft they furnish a response to a stimulus, the more firmly that response doth become connected to that selfsame stimulus.

In a typical experimental arrangement, a feline is ensconced within a cage. The feline may unlock an escape hatch by the pushing of a stick or the pulling of a chain. After a sequence of random responses, the feline eventually escapes by the rendering of a response that unlocks the hatch. The feline is then returned to the cage. Over successive trials, the feline doth reach the desired goal (escape) more promptly and commits fewer errors prior to responding correctly.

Trial-and-error instruction occurs gradually (incrementally) as successful responses art established and unsuccessful ones art abandoned. Connections art formed mechanically through repetition; conscious awareness is not a necessary prerequisite. Animals do not 'catch on' or 'have insight'. Thorndike did comprehend that human instruction is of greater complexity, inasmuch as persons engage in other forms of instruction involving the connecting of ideas, analysing, and reasoning (Thorndike, 1913b). Nonetheless, the similarity in research results from studies involving both animal and human subjects did induce Thorndike to expound complex instruction via elementary instructional principles. An educated adult doth possess millions of stimulus-response connections.

Thorndike's fundamental notions concerning the acquisition of knowledge are encapsulated within the Laws of Exercise and Effect. The Law of Exercise comprises two distinct components: Firstly, the Law of Use, which posits that the bond between a stimulus and its corresponding response is fortified through repeated engagement. Secondly, the Law of Disuse, which asserts that when a stimulus fails to elicit a response, the strength of their connection diminishes, eventually leading to its obliteration, or forgetting. The temporal duration elapsing before a response is executed exerts a direct influence on the extent to which the connection's potency wanes; ergo, the longer the interval, the greater the decline.

The Law of Effect constitutes a cornerstone of Thorndike's theoretical framework (Thorndike, 1913b):

When a modifiable connection between a situation and a response is established, and this connection is either accompanied or followed by a state of affairs deemed satisfactory, the strength of that connection is augmented. Conversely, should the connection be accompanied or succeeded by a state of affairs perceived as vexatious, its strength is diminished.

Responses that yield gratifying (rewarding) outcomes are readily assimilated, whereas those engendering irksome (punishing) consequences are not. This construal of learning is inherently functional, insofar as satisfiers, that is, responses that precipitate desirable outcomes, empower individuals to adapt effectively to their surrounding environments.

The ensuing study serves as an exemplification of the Law of Effect in practical application (Thorndike, 1927). Participants were presented with a series of 50 paper strips, varying in length from 3 to 27 centimetres (cm), shown individually. Adjacent to each strip was another, invariably measuring 10 cm in length, and participants were apprised of this fact. Initially, they were tasked with estimating the length of each strip without the provision of feedback. Subsequently, following this preliminary assessment, the selfsame 50 strips were re-presented, seriatim. After each estimation, the experimenter conveyed whether their judgment was 'correct' or 'incorrect'. This iterative process was sustained across several days, whereupon participants were once more invited to estimate the lengths of the strips, sans any feedback regarding accuracy. Post-training, it was observed that participants' length estimates bore a closer resemblance to the actual lengths of the strips than had their prior estimations. Thorndike inferred from these findings, analogous to those derived from experiments wherein animals were rewarded with victuals or liberty, support for the proposition that gratifying (correct) stimulus–response associations are fortified, whilst vexatious (incorrect) ones are weakened.

Thorndike's (1913b) theory encompassed sundry other principles of relevance to education. One such principle is the Law of Readiness, which posits that when one is prepared (ready) to act, to do so is rewarding, and failure to do so is punishing. If one experiences hunger, responses conducing to the procurement of sustenance are in a state of readiness, whereas responses not conducing to such are not. If one is fatigued, to be compelled to exercise is a punishment. Applying this notion to pedagogy, we might aver that when students are in a state of readiness to assimilate a particular action (in terms of developmental stage or prior skill acquisition), then behaviours that foster this assimilation shall be rewarding. When students are not ready to learn, or do not possess the requisite prerequisite skills, then attempting to learn is both punishing and a profligate waste of time.

The principle of associative shifting pertains to a situation wherein responses elicited by a particular stimulus are, in due course, elicited by an entirely disparate stimulus if, upon repeated trials, there occur small alterations in the nature of the stimulus. For example, to instruct students in the division of a two-digit number into a four-digit number, we should first instruct them in the division of a one-digit number into a one-digit number, and thereafter gradually augment the number of digits in the divisor and dividend.

The principle of identical elements affects transfer (generalisation), or the degree to which the strengthening or weakening of one connection engenders a similar change in another connection (Hilgard, 1996; Thorndike, 1913b; vide Chapter 7). Transfer ensues when situations evince identical elements and necessitate similar responses. Thorndike and Woodworth (1901) discovered that practice or training in a skill in a specific context did not ameliorate one's capacity to execute that skill generally. Thus, training in the estimation of the area of rectangles does not advance learners' ability to estimate the areas of triangles, circles, and irregular figures. Skills ought to be taught with differing types of educational content, so that students might comprehend how to apply them.

Thorndike, subsequent to further investigation, revised his Laws of Exercise and Effect, finding them unsupported by empirical evidence (Thorndike, 1932). He relinquished the Law of Exercise, observing that mere repetition of a given circumstance did not invariably serve to “stamp in” corresponding responses. In one particular experiment, participants were tasked with drawing lines, whilst deprived of visual feedback, purporting to be 2, 4, 6, and 8 inches in length, repeated hundreds of times over several days (Thorndike, 1932). Were the Law of Exercise veracious, the response most frequently executed during the initial hundred or so drawings should have exhibited an augmented prevalence thereafter; however, Thorndike discerned no substantiation for this hypothesis. Instead, mean lengths varied over time, suggesting an experimental approach to determining the correct length, arising from uncertainty. Thus, the recurrence of a circumstance may not necessarily amplify the future likelihood of an identical response.

With regard to the Law of Effect, Thorndike initially posited that the consequences of satisfiers (rewards) and annoyers (punishments) were opposing yet commensurate; however, research demonstrated this to be fallacious. Rather, rewards fortified connections, whereas punishment did not invariably attenuate them (Thorndike, 1932). Instead, connections are weakened when alternative connections are strengthened. In one such study (Thorndike, 1932), participants were presented with uncommon English words (e.g., edacious, eidolon), each followed by five common English words, one of which constituted a correct synonym. On each trial, participants selected and underlined a synonym, whereupon the experimenter would pronounce “right” (reward) or “wrong” (punishment). Reward augmented learning, whilst punishment did not diminish the probability of the response occurring to the presented stimulus word.

Punishment serves to suppress responses, albeit without eradicating them from memory. Punishment is an ineffectual means of modifying behaviour, insofar as it fails to impart correct behaviours, instead merely informing subjects of proscribed actions. This tenet holds true with regard to cognitive skills. Brown and Burton (1978) observed that students acquire buggy algorithms (incorrect rules) for problem-solving (e.g., subtract the smaller number from the larger, column by column, 4371 - 2748 = 2437). Whilst students may be apprised of the method's incorrectness, and furnished with corrective feedback and practice in correct problem-solving, they assimilate the correct method without effacing the erroneous one.

As a Professor of Education at Teachers College, Columbia University, Thorndike penned numerous volumes addressing topics such as educational aims, learning processes, pedagogical methods, curricular sequences, and techniques for assessing educational outcomes (Hilgard, 1996; Mayer, 2003; Thorndike, 1906, 1912; Thorndike & Gates, 1929). A selection of Thorndike’s manifold contributions to education are delineated hereunder.

Principles of Teaching

Instructors ought to aid pupils in the establishment of salutary habits. As Thorndike (1912) duly noted:

• Form habits with diligence, eschewing the expectation of their spontaneous creation.
• Exercise caution in the formation of any habit destined for future dissolution.
• Abstain from the simultaneous formation of multiple habits when a singular one suffices.
• Ceteris paribus, ensure the habit is formed in the manner of its intended application.

The final principle cautions against imparting content divorced from its practical applications: “Since the forms of adjectives in German or Latin are invariably employed with nouns, their study should be undertaken in conjunction with nouns” (p. 174). It behoves pupils to comprehend the application of acquired knowledge and skills, with practical usage being assimilated in tandem with the theoretical content.

Sequence of Curricula

A skill should be introduced (Thorndike & Gates, 1929):

• At or proximate to the juncture when it can be employed in a manner that is serviceable.
• When the learner evinces a consciousness of its necessity as a means of fulfilling a useful purpose.
• When its degree of difficulty is optimally suited to the learner’s aptitude.
• When it harmonises most comprehensively with the prevailing level and type of emotions, proclivities, instinctive and volitional dispositions.
• When it is most fully facilitated by immediately preceding learnings and when it shall most fully facilitate learnings which are to follow shortly.

These principles stand in contrast to the conventional placement of content within schools, wherein subjects are segregated (e.g., social studies, mathematics, science). However, Thorndike and Gates (1929) advocated for the teaching of knowledge and skills across diverse subjects. For instance, forms of governance are pertinent not solely to civics and history, but also to English (how governments are reflected in literature) and foreign language (governmental structures in other countries).

Mental Discipline

Mental discipline constitutes the belief that the study of certain subjects (e.g., the classics, mathematics) enhances general mental functioning to a greater extent than the study of others. This was a prevalent view among educators during Thorndike’s time. He tested this notion with 8,500 pupils in grades 9 to 11 (Thorndike, 1924). The pupils underwent intelligence tests a year apart, and their curricula during that annum were compared to ascertain whether particular courses were associated with more substantial intellectual gains. The findings furnished no support for the concept of mental discipline. Pupils possessing greater initial aptitude demonstrated the most significant progress, irrespective of their course of study.

Thorndike (1924, p. 95):

Had our inquiry been conducted by a psychologist from Mars, wholly unacquainted with the theories of mental discipline, and merely attempting to answer the query, “What are the magnitudes of influence of sex, race, age, degree of aptitude, and courses undertaken, upon the gain made during the year in the capacity to think, or intellect, or whatever our extant intelligence tests measure,” he might even dismiss “courses undertaken” with the observation, “The differences are so trifling, and the unreliabilities relatively so substantial, that this factor appears inconsequential.” The solitary causal factor he would assuredly identify as operative would be the intellect already present. Those with the greatest initial endowment manifest the greatest gain during the year.

Thus, rather than presuming that certain subject areas enhance pupils’ mental abilities more effectively than others, we ought to evaluate how disparate subject areas affect pupils’ capacity for thought, as well as other outcomes (e.g., interests, goals). Thorndike’s influential research impelled educators to redesign curricula, moving away from the paradigm of mental discipline.