The Puzzle Box Experiment (Law of Effect) – Edward Thorndike

At the close of the nineteenth century, psychology stood at a precarious crossroads between speculative mental philosophy and rigorous natural science. For centuries, inquiries into animal behavior and human intellect were dominated by anecdotal accounts, anthropomorphic attributions of conscious reason, and theological assumptions regarding the uniqueness of the human soul. The prevailing Victorian paradigm, epitomized by the early work of Charles Darwin and his disciple George John Romanes, sought to demonstrate evolutionary continuity by projecting sophisticated mental faculties—such as conceptual reasoning, empathy, and deliberate foresight—onto non-human species. However, these naturalistic observations lacked systemic experimental control, quantifiable metrics, and reproducible parameters, leaving comparative psychology vulnerable to charges of sentimentalism and subjective bias.

In 1898, an ambitious young researcher at Columbia University named Edward Lee Thorndike published a doctoral dissertation titled “Animal Intelligence: An Experimental Study of the Associative Processes in Animals.” Thorndike’s monograph fundamentally demolished the anecdotal tradition and established modern comparative psychology upon an empirical foundation. By placing hungry animals—most famously domestic cats—inside meticulously crafted, mechanically latched wooden enclosures known as puzzle boxes, Thorndike bypassed the subjective interpretation of animal minds. Instead, he systematically recorded their overt motor actions, physical escape latencies, and gradual shifts in behavioral efficiency across successive trials.

The theoretical cornerstone to emerge from these empirical investigations was the Law of Effect, a functional principle asserting that behaviors followed by satisfying consequences are stamped into the nervous system, while actions yielding discomfort are systematically eradicated. This foundational insight not only inaugurated the school of thought known as connectionism, but it also served as the essential mechanical antecedent to B.F. Skinner’s radical behaviorism, contemporary learning theory, neurobiological models of synaptic plasticity, and modern artificial intelligence algorithms. This treatise provides an exhaustive examination of Thorndike’s seminal apparatus, empirical protocols, theoretical formulations, historical controversies, and enduring epistemological legacy across the behavioral sciences.

1. Historical Context and the Genesis of Comparative Psychology

1.1 Pre-Thorndikian Views on Animal Intelligence

Prior to Thorndike’s empirical interventions, the late nineteenth-century biological literature was permeated by what modern epistemologists categorize as the anecdotal tradition of animal mind analysis. Spearheaded by figures such as George John Romanes, author of the influential 1882 volume Animal Intelligence, naturalists compiled uncritical, second-hand testimonials from pet owners, hunters, and amateur observers. These accounts routinely ascribed complex ratiocination, abstract planning, moral agency, and deductive inference to non-human species. A dog that managed to open a garden gate was presumed to have deduced the mechanical operation of the latch; a cat that retrieved a kitten from danger was hailed for its maternal empathy and philosophical valuation of life. Romanes operated under the evolutionary imperative to demonstrate phylogenetic continuity between animals and humans, yet his methodology relied heavily on anthropomorphic projection, conflating the external outcome of an animal’s action with the internal mental processes that produced it.

This reliance on uncontrolled, anthropocentric interpretation created a crisis of validity within comparative biology. Naturalists operated without baseline controls, blinded to the underlying biological drives, developmental histories, or unrecorded failures of their subjects. In contrast to this mentalistic perspective stood the historical legacy of René Descartes’s mechanistic physiology, which posited that non-human animals were mere automata—complex biological clocks devoid of true consciousness, subjective feeling, or cognitive plasticity. The intellectual landscape was thus polarized between an ungrounded Cartesian mechanism that denied animals even basic perceptual learning and a sentimental teleology that populated the natural world with miniature human minds encased in fur and feathers. The missing dimension was a standardized, reproducible laboratory methodology capable of disconfirming subjective hypotheses through quantifiable behavioral measurement.

1.2 The Rise of Experimental Naturalism and Morgan’s Canon

The necessary methodological corrective to Romanes’s anthropomorphic excess arrived through the work of British comparative psychologist C. Lloyd Morgan. In his 1894 treatise, An Introduction to Comparative Psychology, Morgan articulated a critical epistemological heuristic that would become known as Morgan’s Canon: “In no case may we interpret an action as the outcome of the exercise of a higher psychical faculty, if it can be interpreted as the outcome of the exercise of one which stands lower in the psychological scale.” Morgan did not deny the reality of animal consciousness, but he demanded parsimony in scientific explanation. He insisted that complex behavioral feats should be explained by associative habituation, perceptual discrimination, and trial-and-error experience rather than abstract conceptual thought or rational deduction.

Morgan illustrated this parsimonious stance through naturalistic field experiments, famously documenting the behavior of his fox terrier, Toby, who learned to open a latched gate not by intuiting mechanics, but by randomly probing the boundaries of the yard until his snout fortuitously dislodged the bolt. Morgan’s canon exerted a profound epistemological impact on the emergent generation of American experimentalists. It signaled an urgent transition away from passive natural observation and toward controlled, manipulative laboratory environments where the history of an animal’s sensory-motor exposures could be rigorously tracked, manipulated, and mathematically recorded.

1.3 Thorndike’s Academic Trajectory at Harvard and Columbia

Edward Lee Thorndike entered the academic discipline of psychology during this period of methodological transformation. Initially studying literature at Wesleyan University, he encountered William James’s masterwork, The Principles of Psychology (1890), which captivated him and redirected his intellectual ambitions toward Harvard University. At Harvard, Thorndike began his experimental investigations under James’s direct mentorship. Lacking dedicated institutional laboratory facilities for comparative animal research, Thorndike improvised an experimental chamber within the confines of his Cambridge boarding house. When his landlady objected to his maintenance of domestic chickens within her residential quarters, William James sheltered the fledgling researcher by offering the cellar of his personal home for Thorndike’s preliminary maze-learning studies.

In James’s cellar, Thorndike designed rudimentary labyrinthine mazes using stacked books and upright sheets of cardboard, examining how domestic chicks navigated spatial enclosures to attain social companionship and grain. Desiring more robust institutional backing, extensive laboratory equipment, and financial support, Thorndike transferred in 1897 to Columbia University to complete his doctoral training under James McKeen Cattell, one of the pioneers of psychometrics and quantitative anthropometry. Under Cattell’s rigorous, measurement-oriented mentorship, Thorndike translated his preliminary avian findings into an expansive investigation involving mammalian subjects. In the summer of 1898, Thorndike defended his doctoral dissertation, “Animal Intelligence: An Experimental Study of the Associative Processes in Animals,” published as a monograph supplement to The Psychological Review. This work dismantled centuries of unscientific speculation and established the empirical paradigm of animal behaviorism.

2. Edward Lee Thorndike: The Architect of Connectionism

2.1 Philosophical Foundations of Early Functionalism

Thorndike’s intellectual work was embedded within the nascent movement of American functionalism, championed by William James, John Dewey, and James Rowland Angell. Departing from the German structuralist tradition established by Wilhelm Wundt and Edward B. Titchener—which focused on the static, introspective dissection of adult human conscious contents into elementary sensations—functionalism approached psychological processes as dynamic, evolutionary adaptations. Drawing upon Charles Darwin’s theory of natural selection, early functionalists conceptualized consciousness and behavioral tendencies as functional instruments forged to facilitate an organism’s adaptation and survival within complex, shifting environments.

Thorndike embraced this evolutionary pragmatism, but he applied it far more rigorously than his contemporaries. He maintained that if mental operations possess adaptive survival utility, their primary evolutionary manifestation must be external, physical action that alters the organism’s relationship with its environment. The ultimate criterion of mind was not introspective self-awareness, but practical efficacy: the capacity of an organism to modify its behavioral topography in direct response to the pressures, hazards, and caloric resources of the external world. Consequently, Thorndike sought to extract psychology from introspective philosophy and establish it as an objective, empirical discipline rooted in natural law and biological mechanics.

2.2 The Core Tenets of Connectionist Theory

The systemic theoretical architecture devised by Thorndike to account for behavioral modification became known as Connectionism. This framework posited that learning does not consist of the acquisition of abstract cognitive maps, internal representations, or rational concepts, but rather the formation of specific, direct neural bonds between sensory impressions and motor impulses. Thorndike termed these foundational units of behavior Stimulus-Response (S-R) bonds. According to connectionist doctrine, an organism confronted by a given environmental stimulus configuration undergoes an internal, physical alteration whereby a specific motor pathway is physically coupled to that sensory pattern.

Thorndike dismissed the necessity of conscious cognitive mediation in habit acquisition. In his view, animals do not reflect upon the nature of their tasks, nor do they generate internal hypotheses regarding causality. Rather, the formation of an S-R connection is mechanical and direct:

• The external environmental context supplies the sensory Stimulus ($S$).
• The organism emits an overt behavioral Response ($R$).
• The immediate qualitative consequence of that response acts directly upon the central nervous system to strengthen or weaken the neural pathways linking that specific $S$ to that specific $R$.

This formulation reduced the process of habit acquisition to an automatic, biological chain reaction, excluding teleological intent, conscious deliberation, or abstract reasoning from the explanatory paradigm.

2.3 Methodological Departures from Classical Associationism

While Thorndike’s theory bore the imprint of classical philosophical associationism—as articulated across centuries by thinkers like John Locke, David Hume, and David Hartley—it represented an epistemological break from their doctrines. Classical British empiricists had focused exclusively on the internal association of ideas within a disembodied mental space. Hume analyzed how concepts like contiguity, resemblance, and cause-and-effect bound thoughts to other thoughts within the introspective theater of the human mind. The somatic body and physical motor apparatus were treated as peripheral instruments subordinate to ideological operations.

Thorndike inverted this paradigm by anchoring associationism in physical acts and quantifiable motor outputs. He abandoned the association of ideas in favor of the association of muscular movements with sensory perceptions. For Thorndike, the target of scientific inquiry was not the subjective flow of ideas, but the observable behavioral latency: the precisely measured time elapsed between the presentation of an environmental problem and the execution of a successful physical resolution. By prioritizing external somatic acts over unobservable mental constructs, Thorndike established the methodological blueprint that would define twentieth-century experimental psychology.

3. Engineering the Puzzle Box: Apparatus Design and Mechanical Architecture

3.1 Physical Dimensions and Construction Materials

To rigorously test the associative abilities of non-human animals under controlled conditions, Thorndike engineered an array of novel experimental enclosures termed puzzle boxes. Fabricated primarily from discarded wooden packing crates, raw timber slats, wire meshing, and basic mechanical hardware, these boxes were constructed by Thorndike himself in the Columbia laboratories. The typical puzzle box designed for feline subjects measured approximately 20 inches in length, 15 inches in width, and 12 inches in height. These specific dimensions were deliberately chosen to balance two critical environmental variables: providing adequate interior space for an adult domestic cat to turn around, rear up, and extend its paws, while simultaneously restricting comfortable resting positions or passive behavioral loafing.

The structural walls of the enclosures were not solid timber, but were instead composed of vertical wooden slats spaced roughly one to two inches apart, or covered with coarse wire mesh. This open-slat architectural arrangement served vital methodological and ecological functions:

• It maintained continuous cross-ventilation, preventing subject asphyxiation or thermal distress across extended experimental sessions.
• It exposed the animal to constant visual and auditory inputs from the broader laboratory environment.
• Crucially, it allowed the animal to visually inspect and olfactorily detect an external food reinforcer placed just beyond its physical reach, maintaining high appetitive arousal throughout each discrete trial.

3.2 Latching Mechanisms and Release Protocols

The defining technical feature of Thorndike’s puzzle boxes was their mechanical release architecture. Thorndike designed over a dozen distinct experimental boxes, systematically designated by letters from Box A to Box K, alongside specialized compound variations. Each box featured a unique mechanical contraption operating a spring-loaded or counter-weighted exit door. When the cat correctly manipulated the specific internal latching mechanism, the door swung open, permitting the animal to step out and consume a morsel of food positioned on an exterior platform.

The operational difficulty and mechanical morphology of these release devices were calibrated to test the limits of animal associative acquisition:

• Box A: Required the simple pulling of an interior wire loop suspended from the ceiling, which traveled over an overhead pulley system to retract a bolt securing the spring-loaded door.
• Box C: Required the subject to depress a horizontal wooden treadle mounted flat upon the floorboards, which pivoted a fulcrum to slide an external locking pin out of its hasp.
• Box D: Required the animal to swivel a loose, vertical wooden button or thumb-latch mounted against the wall through an arc of 90 degrees.
• Box K (Compound Box): Required the sequential, multi-stage disengagement of three independent latches: the cat had to first pull a hanging string, subsequently depress a floor treadle, and finally slide a physical bolt before the exit door would yield.

Thorndike instituted rigorous mechanical standardization across trials. Pulleys were lubricated, cords were calibrated to maintain consistent tension, and counterweights—often simple metal weights or heavy lead fishing sinkers—were adjusted so that a constant, minimal physical force (typically between 50 to 100 grams of pressure) was required to trigger release. This ensured that an animal’s success was governed solely by the spatial accuracy and topography of its motor outputs, rather than random mechanical jams or fluctuations in the box’s resistance.

3.3 Sensory Stimuli and Environmental Controls

Thorndike recognized that behavioral consistency required strict management of the surrounding laboratory environment. The experimental room was shielded from unexpected auditory disruptions, sudden human movement, and extraneous intrusions that could induce startle responses or freezing behaviors. Lighting conditions were kept uniform across trials to avoid casting erratic, shifting shadows within the slatted wooden interior of the apparatus.

The primary sensory driver was the strategic placement of the appetitive reward. Thorndike positioned a small ceramic dish containing a measured portion of fresh fish or beef liver immediately adjacent to the exterior exit door, precisely positioned where the slats allowed maximum olfactory diffusion and clear, unobstructed line-of-sight. The cat inside could perpetually smell and see the reinforcer, yet was physically barred from accessing it by the latched barrier. By tethering the goal object directly to the mechanical threshold of the box, Thorndike established an environmental arrangement that converted appetitive motivation directly into intensive behavioral interactions with the internal walls and mechanisms of the apparatus.

4. Experimental Methodology and Procedural Protocols

4.1 Subject Selection and Deprivation States

Thorndike’s primary experimental subjects were domestic cats (Felis catus), varying across immature adolescents and fully developed adults. To establish comparative benchmarks, he additionally conducted parallel experimental trials using domestic dogs (primarily young terriers) and chicks. Recognizing that learning cannot be induced in an animal whose biological appetites are entirely sated, Thorndike systematically regulated the subjects’ nutritional intake to maintain an active appetitive deprivation state. Prior to experimental trials, animals were kept in a state of controlled hunger, being fed only small, carefully measured rations following the completion of testing sessions, or receiving their daily caloric requirement entirely via the incremental rewards earned during successful escapes.

Within the ethical and historical context of nineteenth-century science, animal welfare standards were fundamentally distinct from contemporary institutional mandates. While Thorndike did not subject his animals to invasive surgical interventions or severe physiological trauma, the operational paradigm intentionally leveraged persistent appetitive drive and psychological confinement stress. The animals were subjected to significant energetic and psychological arousal, which was deemed essential to activate their native motor repertoires and overcome baseline lethargy. Thorndike carefully documented individual subject temperaments, noting marked differences between hyperactive, high-drive animals and placid, hypo-reactive subjects, acknowledging that baseline physiological temperament profoundly influenced the initial velocity of behavioral output.

4.2 Standardized Trial Sequences

Thorndike established an uncompromising, repeatable procedural sequence for every subject subjected to the puzzle box apparatus. A complete experimental session was structured into discrete trials executed according to the following operational steps:

• The cat was gently lifted from its holding cage and carefully inserted through a top trapdoor into the puzzle box, with the external latching mechanisms fully engaged and the timer set to zero.
• The immediate placement of the animal and closure of the upper hatch marked the formal operational onset of the trial, at which point the experimenter retreated out of direct view to avoid serving as an extraneous social cue.
• The trial terminated the exact millisecond the animal’s motor actions actuated the internal latch, the exit door swung open, and the subject crossed the threshold to reach the exterior platform.
• Upon crossing the threshold, the cat was permitted to consume a tiny, calibrated morsel of food (approximately one bite-sized piece of fish or meat), calculated to sustain appetitive motivation without producing sensory-specific satiety.
• An inter-trial interval (ITI) was enforced, typically lasting between two to five minutes. During this interval, the animal was either confined to a neutral rest box or permitted to briefly settle on the laboratory floor before being picked up and re-inserted into the puzzle box for the subsequent trial.

This rigorous cycle was repeated between 10 to 40 times per day per subject across several consecutive days or weeks, depending on the operational complexity of the specific box mechanism under investigation. This repetitive trial structure generated robust, longitudinal datasets that tracked performance from absolute operational ignorance to fluent, automatic mastery.

4.3 Quantitative Data Recording and Latency Tracking

To eliminate subjective appraisal from the evaluation of animal intelligence, Thorndike adopted a strict, singular primary dependent variable: escape latency. Utilizing a mechanical stopwatch, he timed each trial from the exact second the cat was enclosed within the box until the mechanical actuation of the release mechanism occurred. Thorndike recorded these temporal values manually onto standardized laboratory ledgers, compiling tabular sequences that tracked the performance of each individual animal across every consecutive trial.

In addition to raw time metrics, Thorndike maintained descriptive chronological records noting the specific motor topographies displayed by the subjects. He recorded the frequency and character of extraneous, unproductive behaviors—such as gnawing at the wood, reaching through slats distant from the latch, clawing at ceiling wires, or vocalizing in distress. By charting escape latencies chronologically along the ordinate (y-axis) against the ordinal number of successive trials along the abscissa (x-axis), Thorndike generated the first formal, systematic graphical representations of learning curves in the history of psychology.

5. Empirical Observations: The Dynamics of Trial-and-Error Learning

5.1 Initial Undirected Exploratory Thrashing

When an uninitiated, hungry cat was introduced into a puzzle box for the first time, its behavioral response was not marked by calm spatial inspection, deductive calculation, or mechanical appraisal. Instead, Thorndike observed an immediate, violent display of instinctive, undirected exploratory behavior and panic-induced escape reactions. The animal confronted with confinement and the proximal smell of food emitted an erratic explosion of its native, species-typical motor repertoire.

Thorndike documented that the naive feline typically clawed violently at anything its paws could reach; it bit persistently at the wooden slats; it thrust its nose and legs through every available gap; it scampered frantically across the floor; and it raked the ceiling mesh with its claws. It did not stand back to visually trace the path of the wire rope leading from the release door to the ceiling pulley, nor did it intellectually evaluate the mechanical function of the treadle. The animal’s actions were entirely unstructured with respect to the mechanical release requirements of the apparatus. Its initial behavior was determined solely by evolutionarily pre-programmed responses to physical entrapment and appetitive deprivation.

5.2 Accidental Discovery and Latency Abatement

Amidst this initial storm of non-directed, erratic motor movements, the animal inevitably, by purely mathematical probability, struck or brushed against the specific latching mechanism. In Box A, a thrashing paw would accidentally hook into the hanging loop of wire; in Box C, a desperate backward leap or frantic foot-stamp would depress the floor treadle; in Box D, an errant swipe would dislodge the revolving button. The mechanical lock yielded, the counterweighted door swung open, and the cat bounded through the opening to consume the exterior food reward.

The time required for this initial escape was typically substantial, often ranging from 5 to 15 minutes (300 to 900 seconds), and occasionally longer if the mechanism required a specific motor orientation. When Thorndike immediately returned the cat to the box for Trial 2, he observed a revealing phenomenon: the animal did not display sudden mastery. It did not walk directly to the wire loop or treadle and actuate it calmly. Instead, the cat resumed its frantic, undirected thrashing. It once again bit the bars, clawed the corners, and shoved its snout through the slats. However, across an extended series of successive exposures, a measurable transformation emerged:

Trial Number	Representative Latency (Seconds)	Observed Behavioral Topography
Trial 1	600 s (10.0 min)	Frantic thrashing, indiscriminate clawing, biting slats, accidental loop pull.
Trial 2	450 s (7.5 min)	Generalized clawing, persistent bar biting, eventual accidental contact.
Trial 5	210 s (3.5 min)	Localized clawing concentrated closer to front door, reduced vocalization.
Trial 10	85 s (1.4 min)	Minimal rear thrashing; pawing focused primarily on door and loop area.
Trial 15	25 s	Brief exploratory lunges directly toward latch mechanism followed by actuation.
Trial 20	6 s	Instant, targeted motor strike upon the wire loop immediately upon box entry.

The escape latencies did not drop precipitously in a single, sudden downward plunge; rather, they underwent a gradual, erratic, but unmistakable downward drift. Over the course of dozens of trials, the time required to escape diminished from hundreds of seconds down to a steady, asymptotic baseline of five, four, or even two seconds.

5.3 Elimination of Extraneous Motor Behavior

The core empirical phenomenon documented by Thorndike was the systematic, progressive extinction of extraneous motor behavior. In early trials, the behavioral portfolio exhibited by the subject was cluttered with hundreds of irrelevant somatic acts—scratching the floorboards, biting the ceiling wires, vocalizing, and clawing non-functional corners. With each successive trial, these unsuccessful motor outputs were pruned away, while the successful motor response was executed with increasing temporal priority.

Crucially, this streamlining process occurred gradually and incrementally. Thorndike emphasized that this smooth, continuous latency reduction stood as empirical proof against the hypothesis that animals achieve rational “insight” or mental comprehension. Had the cat suddenly deduced the mechanical connection between the wire loop, the overhead pulley, and the door bolt, its escape latency would have registered an immediate, permanent step-function drop: a high latency on Trial $N$, followed by an instantaneous and permanent collapse to minimal latency on Trial $N+1$ and all subsequent trials. Instead, the learning curves were jagged, noisy, and continuous, reflecting a slow, blind, mechanical pruning of behavioral errors through pure trial-and-error experience.

6. The Formulation of the Law of Effect

6.1 The Original 1898 Dual-Aspect Postulate

Synthesizing his massive corpus of empirical data, Thorndike formulated the theoretical principle that would become the cornerstone of modern behaviorism: the Law of Effect. In his 1898 dissertation and its subsequent expanded 1911 monograph, Thorndike defined this law through a symmetrical, dual-aspect postulate governing the strengthening and weakening of neural connections:

“Of several responses made to the same situation, those which are accompanied or closely followed by satisfaction to the animal will, other things being equal, be more firmly connected with the situation, so that, when it recurs, they will be more likely to recur; those which are accompanied or closely followed by discomfort to the animal will, other things being equal, have their connections with that situation weakened, so that, when it recurs, they will be less likely to recur. The greater the satisfaction or discomfort, the greater the strengthening or weakening of the bond.”

This formulation introduced a radical concept into the natural sciences: the consequences of a physical action operate retroactively upon the behavioral tendencies of the organism. In Thorndike’s view, the outcome of an event determines whether the neural pathway that produced that event will survive or atrophy. He conceptualized this mechanism as a literal, physical “stamping in” of adaptive neural circuits and a “stamping out” of non-adaptive ones, predicting the biological mechanics of synaptic plasticity decades before the emergence of modern neurophysiology.

6.2 Operationalizing ‘Satisfiers’ and ‘Annoyers’

Thorndike anticipated the critical philosophical counter-attack that his terminology—specifically the words “satisfaction” and “discomfort”—threatened to re-introduce subjective, mentalistic, and anthropomorphic concepts back into psychology. To prevent this regression, Thorndike executed a rigorous operational definition of these states, defining them entirely through observable, objective behavioral interactions rather than internal phenomenal feelings:

• A satisfying state of affairs (a satisfier) is one which the animal does nothing to avoid, often doing such things as attain and preserve it.
• An annoying state of affairs (an annoyer) is one which the animal does nothing to preserve, often doing such things as avoid or terminate it.

By defining satisfiers and annoyers exclusively in terms of approach, maintenance, avoidance, and termination behaviors, Thorndike decoupled his theoretical framework from mentalism. The terms did not refer to internal feelings of pleasure or pain, but to quantifiable, functional relations between an organism’s motor outputs and its environmental consequences. This operational maneuver protected the Law of Effect from teleological mysticism, establishing it as a purely mechanistic biological law.

6.3 The 1930s Truncation of the Law of Effect

More than three decades after his initial dissertation, Thorndike undertook an extensive reassessment of his theoretical system. Conducting experimental investigations with human subjects involving complex verbal learning, vocabulary acquisition, and multiple-choice association tasks, Thorndike uncovered a surprising empirical asymmetry that forced him to revise his original 1898 formulation. He discovered that while positive outcomes (“satisfiers” or rewards) consistently strengthened the S-R bonds they followed, negative outcomes (“annoyers” or punishments, such as verbal rebukes or mild electrical shocks) did not exert an equivalent, direct weakening effect upon the targeted connections.

In his 1932 volume The Fundamentals of Learning, Thorndike formally introduced the truncated (or revised) Law of Effect. He demonstrated empirically that punishing an incorrect response did not erase that response from the organism’s neural repertoire. Instead, punishment frequently served merely to induce behavioral variation, causing the subject to shift to an alternative action without directly abolishing the original connection. Reward was proven to exercise a direct, positive, biological “stamping in” effect via an automatic neurochemical process Thorndike labeled the confirming reaction. In contrast, punishment operated only indirectly by driving behavioral variability. This theoretical truncation foreshadowed contemporary behavioral principles regarding the profound clinical and pedagogical limitations of aversive control.

7. Subsidiary Principles of Thorndike’s Learning Framework

7.1 The Law of Exercise: Use and Disuse

Alongside the Law of Effect, Thorndike formulated a secondary operational principle known as the Law of Exercise, which comprised two complementary sub-laws:

• The Law of Use: Other things being equal, the more frequently an S-R connection is exercised or enacted, the stronger that connection becomes.
• The Law of Disuse: Other things being equal, when an S-R connection is not practiced or exercised over a period of time, the strength of that connection decays through biological atrophy.

The Law of Exercise was fundamentally an articulation of the classical drill and rote-memorization principle: habit strength was presumed to be a direct function of pure physical repetition. However, following rigorous empirical re-evaluations during the late 1920s and early 1930s, Thorndike carried out a dramatic scientific retreat. In controlled studies where blindfolded human subjects were instructed to repeatedly draw a line exactly four inches long thousands of times without receiving any feedback, Thorndike observed that practice alone produced zero improvement. Pure repetition without knowledge of results did not strengthen the target bond. Consequently, Thorndike officially repudiated the Law of Use, declaring that simple repetition of a connection does not of itself strengthen it. The Law of Exercise was thus relegated to a subordinate status: practice was necessary only to provide opportunities for the Law of Effect to operate.

7.2 The Law of Readiness

To establish the physiological and motivational boundaries within which the Law of Effect functioned, Thorndike formulated the Law of Readiness. This law did not refer to intellectual or developmental maturity; rather, it described the transient physical state of an animal’s neuro-muscular “conduction units” to transmit or block neural impulses. Thorndike operationalized readiness through three physiological scenarios:

• When a conduction unit is ready to conduct, for it to conduct is satisfying to the animal.
• When a conduction unit is ready to conduct, for it not to conduct is annoying to the animal.
• When a conduction unit is not ready to conduct, for it to be forced to conduct is annoying to the animal.

This principle integrated physiological drive states directly into the stimulus-response architecture. A cat that was intensely hungry possessed conduction units primed for feeding and escape behaviors; providing the escape route and food produced immediate satisfaction, stamping in the prior motor movements. Conversely, if a fully sated cat were dropped into the box, or if a hungry cat were physically restrained from executing an activated motor program, profound annoyance ensued, disrupting connection formation. The Law of Readiness thus accounted for the variable motivational dynamics that modulate behavioral conditioning.

7.3 Secondary Characteristics: Prepotency and Associative Shifting

Thorndike augmented his primary learning laws with a battery of secondary principles designed to explain complex behavioral phenomena that simple S-R pairings could not fully capture. Foremost among these was the Law of Partial Activity (or Prepotency of Elements). Thorndike recognized that an organism is not confronted by a single, isolated stimulus, but by an overwhelming mosaic of environmental inputs. This principle posited that animals possess the biological capacity to respond selectively to dominant, prepotent elements of a total stimulus situation while ignoring background noise, thereby laying the groundwork for later empirical studies of selective attention and stimulus control.

Additionally, Thorndike articulated the principle of Associative Shifting, a remarkable conceptual precursor to Ivan Pavlov’s discovery of classical conditioning. Thorndike demonstrated that if a response can be reliably elicited by a specific stimulus $A$, and stimulus $B$ is repeatedly presented concurrently alongside $A$ while $A$ is gradually faded or altered, the response will ultimately become bound directly to stimulus $B$. He demonstrated this by training a cat to stand upright upon the presentation of a piece of fish (Stimulus $A$), progressively accompanying the display with the verbal command “Stand up!” (Stimulus $B$), until the verbal cue alone consistently elicited the physical posture. Furthermore, through his principle of Response by Analogy (generalization), Thorndike explained how organisms navigate novel environmental contexts by drawing upon preexisting S-R bonds that share structural or sensory similarities with familiar situations.

8. Methodological Rigor and Quantitative Measurement

8.1 The Learning Curve as an Objective Instrument

The introduction of the graphical learning curve represents one of Thorndike’s most enduring methodological contributions to the scientific enterprise. Prior to his work, reports of animal behavior were narrative descriptions recounting individual, spectacular accomplishments. Thorndike transformed these subjective descriptions into rigorous, Cartesian coordinate systems where performance was captured as a continuous, mathematical function.

By plotting consecutive trial numbers along the horizontal axis and escape latencies (measured in seconds) along the vertical axis, Thorndike rendered the invisible process of habit acquisition directly visible. The resulting curves revealed two critical empirical realities:

• They demonstrated that learning is an incremental rather than an all-or-none phenomenon; the curves did not drop abruptly from the ceiling to the floor, but exhibited a gradual, downward slope marked by day-to-day fluctuations.
• The presence of these gradual slopes systematically refuted the hypothesis that animals resolve spatial and mechanical problems via sudden conceptual insights or mental illumination.

The quantitative learning curve shifted psychology away from metaphysical debates regarding animal souls and anchored it firmly within mathematical kinetics and objective, graphic analysis.

8.2 Elimination of Experimenter Bias

A fatal flaw of nineteenth-century comparative observation was the confounding influence of experimenter bias and unconscious cuing, a phenomenon dramatically immortalized years later by the case of Clever Hans, the German horse purported to perform complex arithmetic, who was actually reading subtle, involuntary micro-expressions of human onlookers. Thorndike recognized these experimental hazards and engineered his puzzle box paradigms to physically isolate the subject from the investigator.

Once the cat was enclosed and the stopwatch initiated, Thorndike sat completely motionless at an observation station placed several feet away, remaining out of the direct line of sight or adopting an impassive, static physical posture. The mechanical apparatus served as an objective intermediary: the latch either opened or remained locked based entirely on the mechanical interaction between the cat’s paws and the hardware, completely independent of human interpretation, emotional attachment, or observational wish-fulfillment. Furthermore, Thorndike verified the replicability of his findings by cross-testing his paradigms across diverse mammalian and avian species, establishing standardized procedural benchmarks that could be independently confirmed across any comparative laboratory.

8.3 Early Mathematical Modeling of Habit Formation

Thorndike did not limit his analyses to basic descriptive charts; he pioneered the quantitative aggregation and mathematical modeling of behavioral parameters. By calculating mean latencies, computing variance across cohorts of animals, and analyzing the mathematical rate of latency decay across trial blocks, Thorndike constructed the earliest formal quantitative measures of habit strength. He recognized that as learning progressed, not only did latency diminish toward an irreducible physiological reaction-time floor, but the statistical variance between consecutive trials steadily converged toward zero.

These early quantitative methodologies laid the empirical foundations for psychometrics and mathematical learning theories. Thorndike’s relentless dedication to measurement—encapsulated by his famous dictum, “Whatever exists at all exists in some amount, and can be measured”—fundamentally transformed American psychology. His puzzle box datasets demonstrated that dynamic, internal biological adaptations could be subjected to the same mathematical rigor, predictive modeling, and statistical verification that characterized classical physics and physiological chemistry.

9. Academic Controversies and Contemporary Critiques

9.1 The Gestalt Counter-Attacks: Wolfgang Köhler and Insight

Thorndike’s connectionist framework and mechanistic view of learning provoked intense international controversy, most prominently from the emerging school of Gestalt psychology. During the First World War, German psychologist Wolfgang Köhler conducted extensive behavioral investigations with chimpanzees at the Prussian Academy of Sciences research station on Tenerife. In his classic work The Mentality of Apes (1917), Köhler launched a devastating critique of Thorndike’s puzzle box paradigm.

Köhler argued that Thorndike’s experimental design was fundamentally rigged to preclude the manifestation of intelligent behavior. Enclosed within a cramped, dark wooden box, unable to see the external mechanical connections between the cords, pulleys, and external latches, the cat had no epistemological alternative other than to engage in blind, chaotic trial-and-error thrashing. Köhler insisted that Thorndike’s apparatus artificially manufactured the appearance of stupidity:

• Gestalt theorists maintained that true problem-solving does not occur via the blind stamping in of random motor twitches, but through insight (Aha!-Erlebnis)—a sudden restructuring of the animal’s total perceptual field.
• In Köhler’s experiments, apes presented with bananas hung out of reach were provided with movable crates or bamboo poles scattered within their full visual field. Under these conditions, the animals often paused, surveyed the environment, and suddenly stacked the crates or joined the poles to reach the reward.

Köhler asserted that when an animal is granted an unobstructed view of all relevant environmental components, problem-solving manifests as sudden, conscious, insightful restructuring rather than gradual, mechanical S-R bonding.

9.2 Ecological Validity and Biological Constraints

A parallel critique emerged from twentieth-century European ethology, pioneered by Konrad Lorenz and Nikolaas Tinbergen. Ethologists argued that Thorndike’s laboratory setups lacked ecological validity. By trapping animals inside artificial, unnatural wooden cages that bore no structural resemblance to their evolutionary habitats, Thorndike stripped them of their natural environmental adaptations. An animal subjected to severe confinement stress and immediate appetitive starvation will instinctively fire off high-arousal species-specific defense reactions (SSDRs), behaviors that conceal its evolutionary problem-solving capacities.

Decades later, researchers such as Keller and Marian Breland (1961) demonstrated the power of biological constraints on learning through their work on the “misbehavior of organisms.” They proved that animals are biologically prepared by evolutionary natural selection to form certain associations effortlessly, while being biologically counter-prepared to form others. Thorndike’s assumption that any arbitrary stimulus could be linked with equal facility to any arbitrary motor response via the Law of Effect overlooked the powerful influence of innate evolutionary heuristics. Cats are evolutionarily adapted to hunt, pounce, and stalk; they are not biologically engineered to pull artificial wire loops or twist wooden thumb-latches to escape confining crates. Consequently, what Thorndike measured was not necessarily pure intelligence, but an animal’s capacity to overcome innate behavioral programming under conditions of severe confinement stress.

9.3 Watson and Classical Behaviorism

While Thorndike was viewed by European cognitive and ethological theorists as excessively mechanistic, he faced an entirely opposite critique from the radical wing of American behaviorism. John B. Watson, the founder of classical behaviorism who published his manifesto “Psychology as the Behaviorist Views It” in 1913, expressed deep skepticism regarding Thorndike’s theoretical vocabulary. Watson argued that Thorndike had compromised his scientific purity by invoking subjective, unobservable, and mentalistic constructs—most egregiously the concepts of “satisfaction” and “discomfort.”

Watson asserted that postulating internal affective states like “satisfaction” to explain the physical strengthening of an S-R connection was a dangerous concession to subjective psychology. Watson attempted to explain all habit formation exclusively through the objective, non-affective physical principles of frequency and recency:

• The successful response is inevitably the most recent motor act executed before the animal exits the chamber.
• Over dozens of successful trials, that single successful response is repeated on 100% of the completions, giving it an objective mathematical advantage in raw frequency over any single errant response.

Watson insisted that objective behaviorism had to purge Thorndike’s Law of Effect of its affective terminology. Thorndike vigorously defended his law, proving empirically that frequency alone without reward failed to produce robust habit acquisition. This theoretical dispute between Thorndike’s consequence-driven model and Watson’s frequency-driven model ignited decades of debate that culminated in the rise of operant conditioning.

10. Evolution from Thorndike to B.F. Skinner’s Operant Conditioning

10.1 Structural Evolution: Puzzle Box versus Operant Chamber

The conceptual transition from Thorndike’s connectionism to B.F. Skinner’s radical behaviorism represents one of the most transformative methodological evolutions in the history of psychology. While Skinner fully embraced the foundational premise that behavior is shaped and selected by its environmental consequences, he fundamentally redesigned the experimental apparatus, inventing what is popularly termed the Skinner Box (or operant conditioning chamber).

Architectural Parameter	Thorndike’s Puzzle Box (1898)	Skinner’s Operant Chamber (1938)
Experimental Paradigm	Discrete-Trial Procedure: Animal must be physically removed and reinserted after each single trial.	Free-Operant Procedure: Animal remains inside continuously; free to respond repeatedly at its own pace.
Primary Dependent Variable	Escape Latency: Elapsed time (seconds) to achieve mechanical release from the box.	Response Rate: Number of operant actions (lever presses/key pecks) per unit of continuous time.
Data Recording Method	Manual Chronometry: Stopwatches, handwritten logs, and manual Cartesian plotting.	Automated Cumulative Recorder: Continuous electromechanical stylus tracking responses in real time.
Consequence Topography	Negative Reinforcement & Escape: Termination of confinement paired with primary food access.	Positive Reinforcement: Immediate automated delivery of uniform food pellets or water droplets.

The crucial breakthrough of Skinner’s free-operant methodology was that it liberated behavioral research from the disruptive intervention of the experimenter. In Thorndike’s discrete-trial setup, every escape required the experimenter to manually seize the cat, reset the mechanical latch, and re-confine the animal. This handling introduced significant stress and disrupted the natural flow of behavior. In Skinner’s operant chamber, the rat or pigeon remained inside for hours, permitting the automated delivery of hundreds of reinforcements while a mechanical cumulative recorder traced the animal’s continuous, uninterrupted rate of response.

10.2 Conceptual Metamorphosis: S-R Bonds versus Three-Term Contingency

Along with structural apparatus advancements came a profound conceptual shift. Thorndike had conceptualized learning as a mechanistic, rigid Stimulus-Response (S-R) bond, proposing that the external stimulus physically compels the emission of a specific muscular reflex once the connection has been stamped in by reward. Skinner recognized that behavior is rarely a direct, mechanical reflex wired to static sensory inputs. Instead, Skinner introduced the concept of the operant: behavior that is emitted spontaneously by the organism and operates upon the environment to generate consequences.

Skinner replaced Thorndike’s direct S-R bond with the sophisticated Three-Term Contingency:
$$\text{Discriminative Stimulus } (S^D) \rightarrow \text{Operant Response } (R) \rightarrow \text{Reinforcing Stimulus } (S^R)$$
In this framework, the antecedent stimulus ($S^D$) does not rigidly compel or pull the response out of the organism like an automatic reflex; rather, it merely sets the occasion upon which that operant response, if emitted, will produce the reinforcing consequence ($S^R$). Furthermore, Skinner definitively purged Thorndike’s affective vocabulary. He abolished the term “satisfier,” replacing it with the functionally defined reinforcer—any stimulus event that, when presented contingent upon a behavior, systematically increases the future probability of that response class. Skinner simultaneously discarded Thorndike’s hypothetical neurophysiological speculations regarding synaptic stamping-in, advocating instead for a functional, atheoretical analysis of observable behavioral relations.

10.3 Schedules of Reinforcement and Behavioral Plasticity

Thorndike’s puzzle box procedures operated under a rigid, unvarying continuous reinforcement (CRF) schedule: every single successful latch manipulation resulted in immediate escape and food delivery. Thorndike did not investigate what occurred when consequences were delivered intermittently, probabilistically, or non-contingently. Skinner dramatically expanded the boundaries of behavioral science by uncovering the dynamics of intermittent schedules of reinforcement.

Skinner demonstrated that behavior maintained on partial reinforcement schedules—such as Fixed Ratio (FR), Variable Ratio (VR), Fixed Interval (FI), and Variable Interval (VI)—generates profoundly different, highly stable rates of responding that are extraordinarily resistant to extinction. A rat trained to press a lever on a Variable Ratio schedule will emit thousands of responses without a single reinforcer, mirroring the persistent behavioral patterns observed in human gambling. By moving beyond Thorndike’s basic continuous schedules, Skinner established how subtle environmental contingencies generate, maintain, and extinguish behavioral plasticity across all animal taxa.

11. Pedagogical and Practical Applications of the Law of Effect

11.1 Classroom Design and Curriculum Engineering

Following the completion of his comparative animal investigations, Thorndike transitioned to Teachers College, Columbia University, where he served as a foundational architect of modern educational psychology. Applying the principles derived directly from his puzzle boxes to human learning, Thorndike revolutionized classroom design, textbook layout, and pedagogical instruction. He launched a systematic campaign against the classical educational doctrine of “formal discipline”—the long-held belief that subjecting students to difficult, abstract subjects like Latin and Euclidean geometry acted as a general mental muscle-builder, improving overall intellect.

Thorndike proved empirically that learning is highly specific. Through his Identical Elements Theory of transfer, he demonstrated that training in one cognitive domain improves performance in another only to the precise degree that the two domains share identical S-R components. Consequently, Thorndike completely redesigned curricula to prioritize practical, sequenced, and directly applicable skills:

• In reading instruction, he conducted massive statistical analyses of English literature to determine word usage frequencies, publishing The Teacher’s Word Book (1921) to ensure vocabulary was introduced in sequenced, manageable tiers.
• In mathematics education, he replaced convoluted, rote-memorization arithmetic puzzles with sequenced, real-world problems that provided immediate confirmation and tangible feedback.
• He designed early structured workbooks containing self-checking answer keys, providing an immediate satisfying state of affairs that stamped in correct cognitive associations, directly anticipating the programmed learning machines and educational software of the late twentieth century.

11.2 Behavior Modification in Applied Psychology

The translation of the Law of Effect into clinical, institutional, and industrial contexts gave rise to the discipline of Applied Behavior Analysis (ABA) and contemporary behavior modification. In mental health clinics, psychiatric facilities, and educational programs for neurodivergent populations, practitioners operationalized Thorndikian consequences through the implementation of token economies. In these systems, adaptive, functional behaviors—such as self-care, social communication, and task completion—are immediately followed by the delivery of secondary, exchangeable reinforcers (tokens), precisely mirroring the appetitive mechanics that guided Thorndike’s cats to their food dish.

In industrial and organizational settings, Thorndike’s principles catalyzed the development of Organizational Behavior Management (OBM). By structuring performance metrics, aligning corporate incentive programs, and providing rapid, contingent feedback for operational goals, corporations systematically stamped in high-productivity behaviors while phasing out inefficient organizational habits. In modern clinical psychology, evidence-based interventions for anxiety disorders—such as exposure therapy and systematic desensitization—rely directly on Thorndikian extinction protocols, exposing the patient to anxiety-provoking stimuli in the absence of aversive outcomes, thereby dismantling entrenched, maladaptive avoidance loops.

11.3 Animal Training and Ethological Husbandry

Few domains have been as profoundly transformed by Thorndike’s Law of Effect as the fields of commercial animal training, zoological management, and domestic pet husbandry. For centuries, traditional animal training relied heavily on aversive control, physical coercion, pain-compliance tools, and negative punishment. Animals were subjected to choke chains, whips, spurs, and shock devices to force compliance through the suppression of non-compliant behaviors.

The operational insights derived from the puzzle box fostered a global paradigm shift toward positive-reinforcement methodologies, exemplified by modern clicker training:

• Developed by operant theorists and popular animal trainers such as Karen Pryor, clicker training utilizes a distinct auditory cue (the click) as a conditioned, secondary reinforcer paired with food.
• The click functions with microsecond precision to mark the exact physical motor movement the trainer seeks to strengthen, delivering an immediate “confirming reaction” that stamps in the target behavior.
• In contemporary zoological facilities, this methodology enables modern ethological husbandry: dangerous apex predators, marine mammals, and massive pachyderms are routinely trained to voluntarily present paws for veterinary inspection, accept blood draws, and enter transport crates entirely through voluntary positive-reinforcement contingencies.

This operational transition has eliminated coercive physical restraint across modern facilities, transforming animal welfare standards worldwide.

12. Epistemological Legacy and Impact on Cognitive Science and Artificial Intelligence

12.1 Neural Substrates of the Law of Effect

While Edward Thorndike’s original late nineteenth-century hypotheses regarding the biological “stamping in” of neural connections were speculative, modern neurobiology has validated his functional assertions with remarkable biological precision. In 1949, Canadian neuropsychologist Donald O. Hebb published The Organization of Behavior, articulating his famous neurophysiological postulate: when an axon of cell $A$ is near enough to excite cell $B$ and repeatedly or persistently takes part in firing it, some growth process or metabolic change takes place in one or both cells such that $A$‘s efficiency in firing $B$ is increased. This rule—popularized as “cells that fire together, wire together”—provided the cellular mechanism for Thorndike’s S-R bonds.

Decades later, the discovery of Long-Term Potentiation (LTP) in the mammalian hippocampus by Terje Lømo and Timothy Bliss proved that physical synaptic conductance is permanently modulated by high-frequency stimulation. Furthermore, contemporary neurobiology has pinpointed the precise chemical substrate of Thorndike’s “satisfier”:

• The mesolimbic and mesocortical dopamine pathways—originating in the ventral tegmental area (VTA) and projecting to the nucleus accumbens and striatum—operate as a direct, biological reinforcement engine.
• Pioneering neurophysiological studies by Wolfram Schultz and colleagues demonstrated that midbrain dopamine neurons fire intensely in response to unexpected rewards, encoding a reward prediction error (RPE).
• When an animal’s action yields an outcome superior to its current expectation, the phasic burst of dopamine retroactively alters synaptic weights within the corticostriatal circuits mediating that motor response, physically stamping in the motor pattern precisely as Thorndike predicted in 1898.

12.2 Reinforcement Learning in Computational Systems

The direct computational realization of Thorndike’s Law of Effect has materialized within modern computer science and artificial intelligence through the subfield of Reinforcement Learning (RL). In their seminal work, Reinforcement Learning: An Introduction, computer scientists Richard S. Sutton and Andrew G. Barto explicitly credit Edward Thorndike as the intellectual grandfather of the entire computational discipline. Sutton and Barto established that the core mathematical problem of artificial intelligence—how an artificial agent can learn to map environmental states onto actions to maximize a numerical reward signal—is the computational formalization of the Law of Effect.

Every core operational problem identified by Thorndike in his puzzle boxes possesses a direct algorithmic analogue within modern machine learning architectures:

• The foundational exploration versus exploitation trade-off directly models the dynamics of Thorndike’s cats: the computational agent must continuously balance the exploration of novel, untested actions (analogous to the cat’s initial undirected thrashing) against the exploitation of known, high-reward actions (the targeted striking of the release latch).
• Algorithms such as Q-learning, SARSA, and Temporal Difference (TD) learning utilize mathematical updates that alter the numerical “value” of a state-action pair based entirely on the scalar consequence that immediately follows the execution of that action.
• Modern Deep Reinforcement Learning frameworks—such as DeepMind’s AlphaGo and AlphaZero—couple deep neural networks with Thorndikian trial-and-error mechanics, executing billions of simulated interactions to independently discover superhuman strategies in chess, Go, complex video games, and autonomous robotic navigation without human intervention.

12.3 Historical Appraisal of Thorndike’s Place in Science

Across the historical evolution of scientific psychology, Edward Lee Thorndike occupies a towering, foundational status. Positioned at the critical historical nexus between William James’s philosophical functionalism and the uncompromising behavioral revolution spearheaded by John B. Watson and B.F. Skinner, Thorndike forged the empirical bridge that transported psychology out of armchair philosophy and into the physical laboratory. His doctoral dissertation, conceived in boarding house cellars and completed in Columbia’s basements, systematically dismantled the anthropomorphic mythology of the Victorian era, demonstrating that complex behavioral adaptations can be unraveled through rigorous, mechanical parsimony.

While later schools of thought rightfully challenged the limits of his models—Gestalt psychology demonstrated the reality of visual restructuring, ethology highlighted evolutionary constraints, and Skinner dismantled the rigid S-R reflex construct—Thorndike’s central thesis remains unshakable. The Law of Effect stands as one of the few truly universal principles in the behavioral sciences, on par with natural selection in biology. From the synaptic plasticity of cortical neurons to the reward prediction error calculations of midbrain circuits, from elementary school pedagogical design to the deep reinforcement learning algorithms driving the modern computational revolution, Thorndike’s 1898 puzzle box experiments continue to serve as the foundational architecture for our scientific understanding of how biological and artificial entities learn from the consequences of their actions.

In the final assessment, the hungry feline scratching frantically within the slatted confines of Box A did more than escape an improvised wooden crate. Through its frantic thrashing, its fortuitous paw-strike upon the suspended wire loop, and its gradual, stuttering convergence upon behavioral efficiency, that animal provided the empirical foundation for a new science of mind and action. Thorndike demonstrated that learning is not an ethereal, divine spark of disembodied reason, but an evolutionary, biological process governed by universal natural laws. In doing so, he forever stamped the Law of Effect into the permanent scientific ledger of human knowledge.

References

• Bliss, T. V., & Lømo, T. (1973). Long-lasting potentiation of synaptic transmission in the dentate area of the anaesthetized rabbit following stimulation of the perforant path. The Journal of Physiology, 232(2), 331–356. https://doi.org/10.1113/jphysiol.1973.sp010273
• Breland, K., & Breland, M. (1961). The misbehavior of organisms. American Psychologist, 16(11), 681–684. https://doi.org/10.1037/h0040090
• Darwin, C. (1871). The Descent of Man, and Selection in Relation to Sex. John Murray.
• Hebb, D. O. (1949). The Organization of Behavior: A Neuropsychological Theory. John Wiley & Sons.
• James, W. (1890). The Principles of Psychology. Henry Holt and Company.
• Köhler, W. (1925). The Mentality of Apes (E. Winter, Trans.). Harcourt, Brace & World. (Original work published 1917).
• Morgan, C. L. (1894). An Introduction to Comparative Psychology. Walter Scott Publishing.
• Pryor, K. (1999). Don’t Shoot the Dog! The New Art of Teaching and Training. Bantam Books.
• Romanes, G. J. (1882). Animal Intelligence. Kegan Paul, Trench, & Co.
• Schultz, W., Dayan, P., & Montague, P. R. (1997). A neural substrate of prediction and reward. Science, 275(5306), 1593–1599. https://doi.org/10.1126/science.275.5306.1593
• Skinner, B. F. (1938). The Behavior of Organisms: An Experimental Analysis. Appleton-Century-Crofts.
• Skinner, B. F. (1953). Science and Human Behavior. Macmillan.
• Sutton, R. S., & Barto, A. G. (2018). Reinforcement Learning: An Introduction (2nd ed.). The MIT Press. https://mitpress.mit.edu/9780262039246/reinforcement-learning/
• Thorndike, E. L. (1898). Animal intelligence: An experimental study of the associative processes in animals. The Psychological Review: Monograph Supplements, 2(4), i–109. https://doi.org/10.1037/h0092987
• Thorndike, E. L. (1911). Animal Intelligence: Experimental Studies. The Macmillan Company.
• Thorndike, E. L. (1921). The Teacher’s Word Book. Teachers College, Columbia University.
• Thorndike, E. L. (1932). The Fundamentals of Learning. Teachers College, Columbia University.
• Watson, J. B. (1913). Psychology as the behaviorist views it. Psychological Review, 20(2), 158–177. https://doi.org/10.1037/h0074428