The history of comparative psychology follows attempts to understand behaviour and mental processes by studying animals, including comparisons with humans. Its central difficulty is that an action can be observed while the process producing it must be inferred. Dependable understanding therefore requires more than a striking animal story: it needs observations and comparisons that distinguish plausible explanations.
This guide follows selected methodological landmarks, especially C. Lloyd Morgan’s account of sustained observation and Edward Thorndike’s experimental study of learning, before connecting them with ethology and present-day evidence reading. Original fictional datasets show how to interpret learning curves, compare tasks and repair misleading claims. The historical apparatus described here is not a recommendation to experiment on animals; the exercises use supplied or invented records.
Begin with The Gold Standard master, or follow the wider History of Zoology. The Education parent explains why understanding should survive a change of context, while How X Works connects the collection’s mechanism-based guides.
Choose your comparative psychology history route
Chapters 1–3: Questions, observation and anecdotes
Chapters 4–6: Experiments, curves and explanations
Chapters 7–9: Species, settings and experience
Chapters 10–12: Samples, history and transfer
Chapter index
Questions, observation and anecdotes · Chapters 1–3
1. A comparison needs a question before it needs a ranking
Experiments, curves and explanations · Chapters 4–6
4. Thorndike made repeated performance a central record
5. Read a learning curve before naming a learning mechanism
6. The same outward success can depend on different information
Species, settings and experience · Chapters 7–9
7. Fair comparison does not always mean identical equipment
Samples, history and transfer · Chapters 10–12
10. Count individuals as well as trials
11. Historical explanation includes the conditions under which knowledge was made
12. Transfer check: does a new apparatus measure the same ability?
Projects and further reading · Chapters 13–14
13. Build a useful learning project without testing an animal
CHAPTER 1 OF 14 · Questions, observation and anecdotes
1. A comparison needs a question before it needs a ranking
An animal opens a container. The observation is interesting, but it does not yet tell us whether the animal learned a movement through repetition, followed a sensory cue, copied another individual or understood a relationship between parts. Those explanations can predict the same successful action under familiar conditions. Comparative psychology becomes informative when it asks what additional observation would separate them.
Comparison can occur between species, between individuals, between developmental stages or between conditions experienced by the same individual. These are different comparisons. A difference between two individuals is not automatically a species difference, and a difference between two tasks is not automatically a difference in mental capacity. The question determines which contrasts are relevant.
Start with the claim’s actual size
Consider the statement that an animal remembered a location. It could mean that it returned to the place, that it used information retained from an earlier visit, or that it represented the place independently of an available smell. Each version makes a stronger claim. A location choice alone may establish the first while leaving the others open.
A useful reader writes down three things separately: the action recorded, the proposed process and the alternative explanations. This small habit makes historical research easier to understand. An author may report observations carefully but attach a conclusion that the design cannot uniquely establish. Another may recognise the uncertainty and propose a further test. The difference matters more than whether the prose sounds confident.
The history explored here is selective rather than a complete international chronology. It concentrates on a recurring methodological problem: how to move from behaviour to an explanation without letting admiration, dislike or familiarity decide the answer. The broader animal sciences contain many other histories, including anatomy, classification and ecology. Comparative psychology overlaps with them while giving special attention to learning and psychological interpretation.
Avoid turning this subject into a ladder with humans permanently at the top and other animals placed below in a single order. A task measures performance under specified conditions. Before ranking species, ask whether the task offers comparable sensory access, motor possibilities, experience and incentives. Otherwise the comparison may mostly reveal which animal the apparatus was designed around.
CHAPTER 2 OF 14 · Questions, observation and anecdotes
2. Morgan’s preface reveals a field arguing about evidence
In the preface to the revised 1903 edition of An Introduction to Comparative Psychology, C. Lloyd Morgan described his interest in the relationship between human psychology and that of other animals. He also acknowledged predecessors, correspondents and disagreements. The source shows a developing conversation rather than a discipline produced by one isolated founder.
Morgan explained that reports from a correspondent, T. Mann Jones, led him to examine claims about young birds through his own observations. He described some results as repetitions of earlier work and noted differences from other observers’ conclusions. This is an important historical pattern: a familiar claim became a question again when observations did not fit it.
Repetition is not the same as copying an answer
If one observer reports that a young animal responds to a particular event without previous experience, a second observer needs more than the same species name to assess the claim. The conditions of development, opportunities for exposure and definition of the response all matter. A repetition can be informative precisely because it makes those conditions explicit.
Morgan also discussed reports of dogs bringing sticks through railings. He urged readers who witnessed impressive acts to use them as starting points for further observation. The useful distinction is between an event that attracts attention and an investigation that establishes what the event means. A surprising success can suggest a hypothesis without settling it.
Read the source’s vocabulary historically. Expressions about higher animals or psychological scales belong to an older conceptual framework. Explaining their role does not require adopting them as an adequate present-day ranking of species. Likewise, Morgan’s emphasis on human introspection tells us about his proposed route into animal psychology; it is not a requirement that every modern investigation use that approach.
The preface is valuable because it preserves uncertainty and revision. Morgan acknowledged that some terminology and arguments had changed between editions. For a student, that is a reason to record the edition and passage being discussed. A statement attributed simply to Morgan may conceal a difference between versions, or between his own words and a later summary of them.
CHAPTER 3 OF 14 · Questions, observation and anecdotes
3. An anecdote can discover a problem without resolving it
Imagine a fictional report: a dog pushed a gate once, the gate opened, and the observer concluded that the dog understood the latch. Nothing in that description requires the event to be fabricated. The problem is that the observation leaves several explanations compatible with the same outcome. Perhaps the push moved a loosely fitted mechanism. Perhaps a familiar movement happened to work.
A useful investigation would need the preceding history as well as the successful moment. Had the animal encountered the gate before? Did anyone assist? How many unsuccessful attempts occurred? Were similar movements made when the gate did not open? The absence of this information limits interpretation even when everyone involved is honest.
Original calculation: the missing unsuccessful attempts
Suppose an archive contains ten memorable reports of successful solutions. A later notebook reveals that these successes occurred among 200 comparable opportunities. The recorded success fraction is 10/200 = 5%, not ten successes out of ten attempts. The denominator changes what the reports suggest, although it does not explain the successful cases by itself.
Now suppose the 200 opportunities were not comparable. Some involved a different apparatus, and others occurred after extensive practice. Combining them into one percentage may obscure the very process under investigation. The repair is to preserve the conditions and sequence, then calculate summaries for genuinely meaningful groups. A correct division cannot rescue poorly defined categories.
This lesson applies to historical anecdotes about humans too. Memorable performances are more likely to be retold than ordinary failures, so a collection of stories is not automatically a representative sample. That does not make stories worthless. They can identify abilities worth investigating and preserve events that formal institutions ignored. Their evidential role should match their design.
When reading an old animal story, make the smallest defensible claim first: this source reports that an action occurred under these described circumstances. Then separate what remains unknown. You can remain interested in the event without treating a dramatic interpretation as established. Curiosity becomes more productive when it generates a precise next question.
CHAPTER 4 OF 14 · Experiments, curves and explanations
4. Thorndike made repeated performance a central record
Thorndike’s Animal Intelligence, published as a collected volume in 1911, brought together earlier experimental studies. The experimental chapter directly criticised reliance on selected anecdotes, uncertain previous experience and observations that could not be repeated under regulated conditions. It provides a primary source for understanding his methodological ambitions.
He described enclosures from which an animal could escape through an action such as operating a mechanism. He recorded behaviour and the time taken to achieve escape across repeated trials. The historical importance is that the sequence of performance became evidence. Instead of preserving only a successful ending, the record could show how performance changed with experience.
What the apparatus made visible
A sequence can distinguish a slow first success followed by consistently rapid performance from a gradual decrease interrupted by failures. Those patterns invite different explanations. They do not make the explanation automatic. A curve is a description of measured performance; identifying the learning process still requires an argument about conditions and alternatives.
Thorndike expressed strong confidence in the objectivity of timing records. A modern reader should separate the value of a repeatable measure from the stronger idea that measurement is free of judgement. Someone still defines the start and end, decides how to classify an incomplete trial, selects cases and interprets the resulting series. The historical text is evidence of his position, not an instruction to accept every methodological claim in it.
The original procedures also included conditions that should not be copied as casual demonstrations. Reading experimental history does not require reconstructing confinement or deprivation. Students can examine published records, analyse fictional datasets or use non-animal simulations to understand the reasoning. A method’s historical influence and its suitability for use now are separate questions.
Thorndike’s contribution in this account is therefore specific: repeated, documented performance offered a stronger basis for investigating learning than a collection containing only notable successes. The limitation is equally specific: controlled repetition does not by itself establish that a task fairly represents every ability of an animal, or that its explanation transfers unchanged to another species.
CHAPTER 5 OF 14 · Experiments, curves and explanations
5. Read a learning curve before naming a learning mechanism
Consider an invented series of completion times: 120, 90, 65, 50 and 45 seconds across five opportunities on the same task. The overall decline is clear. The final time is 75 seconds shorter than the first, a reduction of 75/120 = 62.5%. That describes performance in this example; it is not a measured result from Thorndike’s experiments.
The change between successive trials is 30, 25, 15 and 5 seconds. Improvement becomes smaller in absolute seconds over these five observations. A reader might describe that pattern as rapid early improvement followed by slower gains. It would be premature to claim a permanent plateau from one final interval, or to infer a unique mental process from the shape alone.
Compare two records with the same endpoints
A second invented individual records 120, 118, 119, 46 and 45 seconds. The first and last values match the first series, but the path differs. Reporting only the 62.5% reduction would conceal that difference. The second record raises questions about a sudden change, an altered strategy, an unnoticed assistance or ordinary variability that a short series cannot resolve.
Neither a smooth curve nor an abrupt one is a direct photograph of thought. A researcher must connect the proposed mechanism to further predictions. Would a change in irrelevant surface features preserve performance? Would a rearrangement of the relevant relationship disrupt it? Does performance remain improved after a delay? The appropriate comparison depends on the claim.
There is also a measurement issue. If a timer records only whole seconds, differences smaller than that resolution cannot be read from the record. If trials stop after a time limit, an unfinished performance is not an exact completion time equal to that limit. Preserve that distinction rather than treating every cell in a table as the same kind of observation.
A careful historical reading keeps the sequence, definitions and recording rules together. A published graph may summarise a rich notebook or conceal missing detail. Ask what the graph enables you to conclude, what its labels leave uncertain, and which part of the author’s interpretation requires evidence beyond the plotted points.
CHAPTER 6 OF 14 · Experiments, curves and explanations
6. The same outward success can depend on different information
An animal repeatedly chooses the correct location. One explanation is memory for place; another is a cue attached to the target. If both cues always point to the same option, the recorded choices cannot distinguish them. The problem is not insufficient enthusiasm or an inadequate label. The competing explanations make the same prediction in the observed situation.
In a fictional paper exercise, imagine a target on the left carrying a visible marker. The record shows eight correct choices out of ten. The observer calls this location memory. But the marker and location were never separated. The proportion 8/10 = 80% accurately describes choices under that arrangement while leaving the claimed mechanism unresolved.
A diagnostic comparison on paper
Draw two hypothetical follow-up conditions. In one, the marker stays on the left. In the other, it moves to the right while the relevant location remains left. A place-based and a marker-based account now predict different choices. This is a reasoning exercise about experimental design, not an instruction to manipulate animals yourself.
Even such a comparison has limits. Moving the marker may create novelty or change attention. A useful design considers those effects rather than assuming that one changed feature has only one consequence. The aim is to build converging evidence, not to invent a single dramatic test that claims to settle every interpretation.
Historical reports often lack the detail required to reconstruct which cues were available. A photograph can reveal the apparatus without showing odours, sounds, prior handling or the observer’s position. An explanation that depends on the absence of a cue needs evidence that the cue was actually excluded. Silence in the report is not the same as a documented control.
For further practice with the distinction between observation and psychological interpretation, use Teaching Psychology: Mind, Behaviour and Ethical Evidence. The general lesson is shared: name the process only as confidently as the design allows, and state which alternative remains plausible.
CHAPTER 7 OF 14 · Species, settings and experience
7. Fair comparison does not always mean identical equipment
Two species receive the same task and produce different scores. It is tempting to call the procedure fair because the equipment was identical. But equal physical presentation does not guarantee equal access. Animals can differ in the information they detect, the movements they can make comfortably and the conditions under which they engage with a task.
A purely visual task may investigate a meaningful ability in one context while poorly representing an animal whose ordinary information use is organised differently. A task requiring a particular manipulation may mix learning with motor demands. Before interpreting an apparent difference in cognition, identify the entire route from stimulus detection to recorded response.
Original case: an accessible response changes the interpretation
Suppose fictional Species A completes a button task on 18 of 20 opportunities and Species B on 8 of 20. The observed proportions are 90% and 40%. These data establish a difference on the task as presented. They do not establish a general 50-percentage-point difference in intelligence.
Now suppose independent observations show that the response mechanism is difficult for B to operate. A redesigned response could investigate whether the earlier gap depended on that requirement. If performance changes, the result informs the interpretation of the task. It does not prove that every difference between the species disappears, nor does it justify retroactively replacing the original recorded values.
Comparability is an argument about what is held equivalent and why. Sometimes identical apparatus is valuable because it fixes a physical condition. Sometimes species-appropriate arrangements are needed to examine a shared functional question. Both choices need justification. The reader should ask which question the comparison can answer rather than declaring one design universally fair.
This is one reason that the history of comparative psychology cannot be reduced to a sequence of increasingly impressive animals. Better methods can alter what counts as a fair opportunity to demonstrate an ability. They can also reveal that an apparently simple ranking combines several abilities that should be investigated separately.
CHAPTER 8 OF 14 · Species, settings and experience
8. Ethology widened the conversation about behaviour
The official 1973 Nobel Prize summary records the joint award to Karl von Frisch, Konrad Lorenz and Nikolaas Tinbergen for research on individual and social behaviour patterns. The award is a historical marker of recognition for behavioural research. It is not a claim that all work on animal behaviour began with those three researchers.
For this guide, the important connection is between investigation of learning under arranged conditions and investigation of behaviour as part of an animal’s life. Questions about what information an animal uses, how a response develops and what circumstances make it useful can connect laboratory and field research. Neither location provides every answer automatically.
What does the setting contribute?
A controlled arrangement can help separate factors that normally occur together. Observation in an ordinary setting can reveal actions, sequences and environmental demands that an artificial task omits. These strengths can be complementary. A field observation can suggest a mechanism to investigate, while an experimental result can suggest which features deserve closer observation outside the apparatus.
Do not turn the history into a simple battle between laboratory workers who studied learning and field observers who studied everything else. Such a contrast is too coarse to evaluate individual studies. Read the actual question, method and evidence. A natural-looking setting can still contain strong experimental control, and a laboratory study can ask a carefully framed ecological question.
The existing Ethology guide develops the biological study of behaviour in more detail. Behavioural Ecology follows the relationship between behavioural decisions, environmental conditions and consequences. These are neighbouring routes, not interchangeable names for every investigation discussed here.
A productive historical account preserves the distinction between institutional recognition and scientific explanation. A prize can document whose work was honoured and for what stated reason. Understanding the evidence still requires the original questions and studies. Prestige can guide a reading route, but it cannot substitute for examining how a conclusion was reached.
CHAPTER 9 OF 14 · Species, settings and experience
9. Development and experience complicate a species label
A species name does not specify an individual’s learning history. Two individuals may have encountered different objects, social partners, tasks or environments before observation began. If the difference matters to the proposed explanation, treating them as interchangeable examples of a species can conceal important information.
Historical records vary in how much background they preserve. A report may describe an animal’s age but say little about earlier exposure. Another may describe previous training but omit how individuals were selected. The appropriate response is to state the limitation, not to invent a standard history for all members of the sample.
A fictional comparison with an overlooked history
Imagine two groups completing a route-choice exercise. Group A makes 16 correct choices in 20 opportunities, while Group B makes 10 in 20. The observed difference is 80% minus 50%, or 30 percentage points. Later, the researcher discovers that A had previous experience with a similar layout and B did not.
The scores remain what they were, but the explanation changes. The result cannot isolate the contribution of the new instruction from prior familiarity. A more informative comparison would account for previous experience through design or analysis, and would report how that information was obtained. Merely adding the word controlled to a study’s description does not solve the problem.
This example also shows why the meaning of naive must be specified. Naive with respect to one apparatus does not mean without relevant experience of any kind. An individual can be unfamiliar with a device while familiar with its materials, response requirements or location cues. The appropriate level of detail depends on the hypothesis.
The historical value of early investigators’ attention to experience lies in making this question visible. The continuing challenge is practical: which experiences matter, which can be documented, and which uncertainty remains? Good comparison turns an unspecified background into an explicit part of the reasoning rather than assuming that the species label settles it.
CHAPTER 10 OF 14 · Samples, history and transfer
10. Count individuals as well as trials
Repeated trials are valuable because they reveal change within an individual. They do not create new independent individuals. A table containing one hundred trials from one animal and a table containing one trial from each of one hundred animals answer different questions, even though each contains one hundred rows.
In the first case, the sequence may reveal how performance changes with experience. In the second, the records may reveal variation across a sampled population under one exposure condition. Neither design is inherently superior. The mistake is to use the number of rows as if it automatically measures the breadth of the sample.
Original calculation: a pooled score hides individual variation
Suppose four fictional individuals each receive ten opportunities and record 10, 9, 3 and 2 successes. The pooled proportion is 24/40 = 60%. Their individual proportions are 100%, 90%, 30% and 20%. The pooled value is useful as a summary of these forty opportunities, but it conceals considerable variation.
If the question concerns whether every individual demonstrated the behaviour reliably, 60% does not answer it. If one individual contributed many more trials than the others, the pooled proportion would also give that individual greater weight. The reader needs both the distribution of trials and the purpose of the summary.
Historical publications sometimes provide detailed records for a small number of animals. Those records can be valuable without representing the entire species. A careful account can say that a pattern occurred in the tested individuals and explain the conditions. Generalisation requires additional evidence about selection, replication and variation.
For a practical reading exercise, mark every unit in a study: species, individual, session and trial. Then locate each reported number at the correct level. This prevents a common mistake in both animal and human research: treating many measurements from a narrow sample as if they were many independent demonstrations across a broad population.
CHAPTER 11 OF 14 · Samples, history and transfer
11. Historical explanation includes the conditions under which knowledge was made
A published experiment is also a record of choices: which species were accessible, which behaviours were measurable, which equipment was available and which questions institutions considered worthwhile. Those choices influence what becomes visible in the literature. A history that lists only conclusions can miss why certain abilities received far more attention than others.
This does not mean that a finding is false because it arose in a particular institution. It means that the scope of a finding must be understood alongside its production. A convenient study population may support an important result while leaving open how widely it generalises. A technique that is easy to quantify may attract research without exhausting the behaviour worth understanding.
Read a methods section as historical evidence
Ask what the researcher could observe directly, what had to be inferred and what was not recorded. Identify the instruments, the operational definitions and the conditions under which records were made. Then distinguish a limitation the author recognised from one a later reader can identify with additional knowledge.
Ethical judgement belongs in this reading. Some historical procedures involved distressing conditions. Their role in the development of a field can be explained without presenting them as acceptable demonstrations. Students can learn the logic from existing records and simulations, and should not use pets or wild animals as subjects for improvised tests.
There is also a question of attribution. Morgan’s preface, for example, names people who supplied observations or criticism. A textbook that retains only the principal author can make the work appear more solitary than the source suggests. Looking at acknowledgements, correspondence and revisions can recover some of the collaborative process without claiming that every contribution is equally documented.
The strongest historical account therefore joins the argument to its setting. What was the problem? What resources made a new observation possible? Which people contributed? What remained uncertain? These questions make scientific development intelligible while avoiding both uncritical celebration and the equally unhelpful assumption that historical limitations erase every useful contribution.
CHAPTER 12 OF 14 · Samples, history and transfer
12. Transfer check: does a new apparatus measure the same ability?
An invented historical report records that 18 of 24 animals completed a task. A modern demonstration records 27 of 30 completing a task with a similar name. Someone concludes that animals have become better at solving the problem. Work through the claim before accepting the comparison.
The historical proportion is 18/24 = 75%. The new proportion is 27/30 = 90%. The observed difference is 15 percentage points. These calculations are correct, but they compare outcomes whose meaning depends on the procedures. The arithmetic alone establishes neither a historical trend nor the cause of the difference.
Change one: response demands
Suppose the older task required a precise manipulation while the new task required only a large movement. The new task may be easier for reasons unrelated to the psychological process of interest. To make the comparison meaningful, the report needs to explain which functional demand is shared and which has changed. Similar names are not evidence of equivalent difficulty.
Change two: exposure and selection
Suppose the older sample included first-time participants, while the new sample contained individuals selected after extensive practice. The observed difference could reflect experience or selection. It cannot be interpreted as a change in a species across historical time without much stronger evidence. The appropriate repair is to describe the samples and restrict the conclusion.
Change three: incomplete records
Now suppose only 20 of the original 24 records survive, and 18 of those show completion. Reporting 18/20 = 90% as the original sample’s completion proportion would ignore four unknown outcomes. Among all 24, the known lower bound is 18/24 = 75%; if all four missing records were successes, the upper bound would be 22/24, approximately 91.7%. These are simple missing-outcome bounds, not confidence intervals.
The defensible answer is that the records describe different observed proportions under incompletely comparable conditions. They may motivate a more careful investigation, but they do not justify the proposed historical trend. Transfer succeeds when the reader carries the reasoning across contexts while reopening assumptions that the new circumstances have changed.
CHAPTER 13 OF 14 · Projects and further reading
13. Build a useful learning project without testing an animal
A good introductory project can use a published historical passage and a fictional data table. Choose one claim about learning, identify the observations offered in support and write down two explanations compatible with them. Then propose what additional kind of evidence would distinguish the explanations. The deliverable is an argument, not a live experiment.
For example, use the two learning sequences in Chapter 5. Plot trial number against completion time, state the observed differences and explain why matching endpoints do not establish matching processes. Add a note about timer resolution and incomplete trials. This joins historical reading with numerical literacy without pretending that the invented values are research findings.
A practical record for each source
Record author, title, edition or date, question, method, observation and interpretation. Keep a separate space for limitations. When quoting an old term, explain its historical role rather than silently treating it as a current category. When paraphrasing a conclusion, preserve the difference between the author’s claim and your assessment of its support.
A second useful activity compares a laboratory question with an ecological question. Ask what each design would need to observe and which conclusion it could support. Use Ethology for behaviour and development, and Behavioural Ecology for environmental conditions and consequences. The aim is to connect perspectives while respecting their different explanatory jobs.
The practical value extends beyond animal research. Headlines about learning technologies, classroom methods and human performance often make the same inferential jumps: a score becomes an ability, a change becomes a cause, or one successful demonstration becomes a universal rule. Comparative psychology’s history supplies concrete ways to notice and repair those jumps.
Finish the project with a changed-context question. If the task, sample or measurement changes, which part of your conclusion survives? A strong answer keeps the actual observation, qualifies the generalisation and identifies the next evidence needed. That is a more useful outcome than memorising a list of famous investigators without understanding their methodological problems.
Is comparative psychology only about comparing different species?
No. Comparisons between species are important, but investigations also examine learning within an individual, differences in experience and responses under changed conditions. The comparison should be chosen to answer a question. Merely placing two species in the same table does not guarantee that the study isolates a meaningful psychological difference.
Does a successful action prove understanding?
It establishes successful performance under the recorded conditions. Understanding is a stronger interpretation whose meaning must be specified. If several processes predict the same success, further evidence is needed. The right response is neither automatic praise nor automatic dismissal, but a comparison that makes the alternatives distinguishable.
Were anecdotes useless to the development of the field?
No. They could draw attention to behaviours worth investigating and preserve observations outside formal institutions. Their limitations concern selection, missing conditions and uncertain prior experience. Morgan’s preface explicitly treats notable observations as starting points for more sustained inquiry. The problem is asking an anecdote to establish more than it can.
Is a controlled experiment free of interpretation?
No. Control can make explanations easier to distinguish, but researchers still define the task, choose measures, classify outcomes and decide how far to generalise. Thorndike’s repeated records were methodologically important without making every inference from them inevitable. A reliable history preserves both the contribution and the limitation.
Which sources should I read first?
Read Morgan’s 1903 preface for the relationship between observation, correspondence and revision. Follow with Thorndike’s 1911 preface and experimental chapter for the role of repeated performance. Use the Nobel Foundation’s 1973 award summary as a concise institutional landmark, then explore the existing Ethology and Behavioural Ecology guides for the neighbouring subject routes.
For the wider history of animal science, return to History of Zoology. For evidence-reading skills across psychology, continue to Teaching Psychology. The Gold Standard master and Education parent connect these lessons to the wider collection.