Purposive Behavior in Animals and Men

Tolman argues that behavior is purposive—directed toward goals—and molar, meaning it should be analyzed at the level of whole acts rather than reduced to reflex arcs or muscle twitches. He rejects Watson's molecular behaviorism, which treated behavior as chains of stimulus-response connections, as well as the introspectionist psychology that invoked consciousness as an explanatory entity. His alternative is a behaviorist system that accommodates purpose and cognition without subjective data. The unit of analysis is the 'behavior act,' characterized by its goal-directedness, and the explanatory constructs are intervening variables defined operationally in terms of observable antecedents and consequences. Tolman calls this approach 'purposive behaviorism' or 'operational behaviorism.'

Intervening variables are hypothetical constructs that mediate between observable environmental conditions (independent variables) and observable behavioral outcomes (dependent variables). They are defined operationally: whatever is learned about behavior through their use must be capable of being expressed in terms of observables. Tolman introduces several such variables, including demand (the drive state), appetite (the specific goal object sought), and sign-Gestalt-expectation (the organism's expectation that a given sign leads to a given goal). Behavior is then a function of these variables combined with stimulus conditions. This framework allowed Tolman to theorize about internal processes while maintaining a behaviorist commitment to operational definition and objectivity.

Latent learning is learning that occurs without reinforcement and is not expressed in performance until a incentive makes it relevant. In the classic experiments, rats were allowed to explore a maze without any food reward. On subsequent trials, when food was introduced in the goal box, these rats performed as well as or better than rats that had been rewarded all along. The unrewarded exploration had produced learning about the maze's layout—a cognitive map—that remained latent until motivation made it performance-relevant. This finding challenged the drive-reduction and reinforcement theories that dominated the era. It implied that reinforcement affects performance rather than learning itself, and it supported the distinction between learning and performance that became central to later cognitive and behavioral theories.

Tolman occupies a middle position between Hull's deductive neo-behaviorism and Skinner's radical behaviorism. Like Hull, he uses intervening variables and formal constructs; unlike Hull, he does not embed them in a deductive mathematical system and he avoids physiological speculation. Like Skinner, he emphasizes molar behavior and the importance of environmental variables; unlike Skinner, he posits internal representations—sign-Gestalt-expectations and cognitive maps—that mediate between stimulus and response. Skinner rejected such constructs as explanatory fictions. The three systems constituted the major alternatives in American learning theory from the 1930s through the 1950s. Tolman's version was often called 'cognitive behaviorism' because of its emphasis on expectation and mental representation.

Sign-Gestalt-expectation is Tolman's term for the organism's learned expectation that a particular sign or cue in the environment will lead to a particular goal. It is not a conscious belief but a behavioral disposition: the animal behaves as if it expects the sign to lead to the goal. Cognitive maps are the broader representations of environmental layouts that organisms acquire through exploration. Together these constructs explain how animals can navigate flexibly, take novel shortcuts, and adjust to changes in the environment—capacities that reflex-arc models cannot easily account for. The concepts were later incorporated into cognitive psychology and animal cognition research, though Tolman's operational definitions were less precise than subsequent formulations.

The book was reviewed as a significant contribution to behavior theory, praised for its rigor and its success in addressing purpose within an objective framework. Some reviewers, particularly those aligned with Hull and Skinner, questioned whether intervening variables like sign-Gestalt-expectation were more than redescriptions of behavior. Others saw Tolman's approach as a bridge between behaviorism and the emerging cognitive psychology. Its long-term influence was substantial: latent learning became a standard topic in learning textbooks, the distinction between learning and performance was widely adopted, and cognitive maps became a foundational concept in cognitive psychology and neuroscience. Tolman's insistence that behavior is molar and purposive also influenced ecological psychology and the study of animal cognition.