The Cognitive Structure of Emotions.

Every emotion you have ever felt was a judgment. Not a sensation, but a judgment. Hold that for a moment. If that's true, it means emotions have a logical structure. And if they have a logical structure, you should be able to map it — precisely enough to predict which emotion a given situation will produce. That is exactly what Ortony, Clore, and Collins set out to do. Their book, The Cognitive Structure of Emotions, builds from a single central claim: emotions are valenced reactions to events, agents, or objects, with their particulars determined by how the eliciting situation is construed. What makes something an emotion is not a raw bodily sensation in isolation. It is the meaning a situation carries for a person, interpreted through their goals, standards, or tastes. The authors call this an appraisal, which is a mental evaluation of whether something is desirable, praiseworthy, or appealing. The result is a theory that treats feeling as a form of reasoning. The framework is organized as what Ortony, Clore, and Collins call successive differentiation. At the very top sits the simplest possible division: positive versus negative valence. From that split, increasingly differentiated emotional states emerge as more information about the situation is processed and incorporated. The authors are careful to say this is a logical structure of the emotion space, a map — rather than a strict timeline of mental steps. And that map has a three-way fork at its heart. The fork works like this. Different kinds of things we encounter are evaluated against different kinds of value structures, producing distinct families of emotion. Events are judged against goals. Agents, which are the people or things that cause things to happen, are judged against standards. Objects and qualities are judged against attitudes. These three value fields — goals, standards, and attitudes — are not interchangeable. They respond to different questions. Did this event help me get what I want? Did this person act the way they should? Do I find this thing appealing? Each branch has its own intensity variable, which is a dial, so to speak, that determines how strongly the emotions in that family are felt. For goal-related emotions, the variable is desirability: how much does this event align with or frustrate what you want? For standards-related emotions, the variable is praiseworthiness: how well does the action meet or violate your moral or social standards? For attitude-related emotions, the variable is appealingness: how much does the object match your tastes? These three variables — desirability, praiseworthiness, and appealingness — are local to their respective branches. Ortony, Clore, and Collins illustrate this with a single clever example. Take a malfunctioning car. If you construe the car as an agent that caused you trouble, the emotion that arises is likely disapproval, which is a standards-related reaction. If you treat the same car simply as an object, the emotion is dislike, which is an attitude-related reaction. Same stimulus. Different appraisal. Different emotion. That's the theory doing real work. The authors also point out an important wrinkle in the attitudes branch. Tastes and attitudes lack the kind of underlying propositional structure that goals and standards possess. You can articulate a goal. You can state a standard. But a taste is harder to reduce to a logical rule, even though tastes are far from simple, serving as class markers and signals of social aspiration. This asymmetry is not a flaw in the theory; it reflects something genuinely true about how different kinds of values operate in human psychology. Now, local intensity variables explain why one fear is stronger than another fear within the same family of emotions. But Ortony, Clore, and Collins go further, identifying four global variables that modulate intensity across all emotion types, regardless of which branch produced them. These are sense of reality, psychological proximity, unexpectedness, and existing level of arousal. These four act as universal amplifiers or dampeners, scaling the strength of any emotion regardless of its origin. The most counterintuitive of these is sense of reality. You might assume that a more real, more immediate tragedy would produce a stronger emotional response. But Ortony, Clore, and Collins link sense of reality to the experience of numbing — the phenomenon where people confronted with enormously tragic circumstances, including losses through death, report a strangely dampened emotional response. The event is objectively catastrophic, but the emotion is muted. That's sense of reality working in reverse, and the theory accounts for it explicitly. The other three global variables — psychological proximity, unexpectedness, and existing arousal — modulate in more intuitive directions: closer, more surprising events in a more aroused state hit harder. With these local and global variables in hand, the authors build out a full taxonomy. Using the process of successive differentiation — first positive or negative valence, then which value field, then which specific appraisal — they specify twenty-two distinct emotion types. Each comes with lexical tokens, which are the everyday words we use to label that emotional state, along with its relevant intensity variables and examples. The authors caution, though, that emotion words are only a shortcut. There is no clean one-to-one correspondence between everyday language and the underlying theoretical space. The word "angry" covers a lot of conceptual territory that the theory carves into distinct states depending on the appraisal that produced it. The directed structure that organizes these twenty-two types has four kinds of internal linkages: sufficiency, necessary, facilitative, and inhibitory. Think of it as a decision tree with richer connective tissue than a simple yes-or-no branch. A sufficiency linkage means a condition is enough on its own to produce an emotion. A necessary linkage means the condition must be present. A facilitative linkage helps produce an emotion without being required. An inhibitory linkage works against one. Together, these four link types let the authors describe why the same underlying valence can branch into joy in one context and relief in another, or into anger in one case and reproach in another. Crucially, this structure is dynamic. Goals can be replaced by new goals. Goal priorities shift. Standards and attitudes evolve. So the tree of feeling is continuously reconfigured as circumstances change and new information arrives. Ortony, Clore, and Collins are not describing a static chart; they’re describing a living architecture that tracks a person's changing relationship to the world around them. That formal, logic-oriented character of the theory is precisely what gave it traction well beyond academic psychology. Ortony, Clore, and Collins explicitly propose that their framework could provide the basis for a set of rules useful in an artificial intelligence system. They name specific applications: natural language comprehension, cooperative problem solving, and planning programs. Because the OCC model makes the antecedents and intensity determinants of each emotion explicit and formalized, it functions as a bridge from cognitive science to computational representations of affect. For a system trying to understand a sentence like "she was devastated when the promotion went to someone else," the OCC framework offers more than a sentiment label. It offers a causal account: an event was appraised against a goal, found to be highly undesirable, and produced a strongly negative goal-related emotion. That kind of structured reasoning about emotional states is something a sentiment classifier trained on word frequencies alone cannot replicate. Mapping the logic of emotion is a step toward systems that don't just detect feeling but understand its source. And perhaps more than the applications, what Ortony, Clore, and Collins leave behind is a way of thinking. Emotions are not noise interrupting rational thought. They are a form of evaluation — structured, differentiable, and in principle as mappable as any other cognitive process. The appraisal is the emotion. That claim, stated with formal precision across twenty-two emotion types and seven intensity variables, is what makes this work still worth reckoning with. This lecture was created by ennepō. Go to https://ennepo.ai to Discover, Create and Follow the latest research in your field. Read when you can. Listen when you want to.