- Identical arousal becomes joy or rage depending on context, showing emotions are constructions, not direct readouts of the body.
- Philosopher Se Yong Bae argues emotions are predictions assembled by the brain's general forecasting system using learned templates called emotion concepts.
- The view reframes anxiety and depression as forecasting gone wrong, with therapy as teaching the brain different bets.
Your Feelings Are Forecasts the Brain Quietly Assembles
The same racing heart can become joy in one room and rage in the next. What your brain does in between is where feeling actually happens.
Key Takeaways
In 1962, the psychologist Stanley Schachter injected dozens of college students with adrenaline and then placed them in different rooms. In one, an actor bounced off the walls in giddy mischief. In another, an actor seethed over an insulting questionnaire. The hormone was identical, the racing hearts identical, yet students in the first room reported feeling happy, and those in the second reported feeling angry. The body said one thing; the mind made it mean two.
This experiment has hung over the science of emotion ever since. If a pounding heart and damp palms can become joy in one room and rage in the next, emotions can’t be simple readouts of what the body is doing. Something else is finishing the sentence.
Writing in the journal Erkenntnis, the philosopher Se Yong Bae of the University of Missouri argues that this “something else” is the brain itself, working as a forecasting machine. Emotions, in his account, aren’t reactions delivered to consciousness after the fact. They’re predictions, assembled in real time from the body’s signals, the situation around them, and a stock of mental templates he calls emotion concepts.
The Body Alone Cannot Say What You Feel
For two decades, the most influential mechanical account of emotion came from the philosopher Jesse Prinz, whose embodied appraisal theory tried to bridge two warring traditions. The somatic camp, going back to William James in 1884, said emotions are perceptions of bodily change. The cognitive camp said emotions are evaluative judgments. Prinz combined them with an elegant analogy. A smoke alarm doesn’t contain the concept of fire, yet it represents fire because it reliably goes off when fire is present. Emotions, he argued, work the same way. The body is the alarm.
The trouble is that the alarms aren’t reliable. A 2018 meta-analysis by Erika Siegel and colleagues reviewed more than 200 studies and found no consistent autonomic signature for fear, anger, or sadness. Heart rate climbs in fear, anger, exertion, and influenza alike. Skin conductance tracks arousal, not category. Facial muscles fire across positive and negative states without a clean signal. The body, it turns out, doesn’t speak the language Prinz needed it to speak.
Prinz himself appears to have come around. In a 2025 paper, he abandoned the detector theory entirely, proposing that emotions are imperatives rather than indicators, commands rather than perceptions of danger or loss. Bae reads this turn as confirmation of the pressure the older view was under, both from philosophical critics and from twenty years of accumulating physiological evidence. Whatever emotions are, they’re not the body keeping watch.
The Brain as a Restless Forecaster
The alternative Bae develops draws on a movement in neuroscience that has gathered force over the past two decades, called predictive processing. Associated with Karl Friston’s free-energy principle and Andy Clark’s writings on the predictive mind, the framework reverses the old picture of perception. The brain isn’t a passive receiver waiting to be told what’s out there. It’s a constant forecaster, generating expectations about what’s about to happen and revising them only when reality contradicts the guess.
The brain on this picture is a Bayesian gambler making bets at every level at once. Lower levels predict edges and sounds; higher levels predict objects and events; the highest predict abstract things like “this is a kitchen” or “this person is angry with me.” When prediction matches input, the system is quiet. When the two diverge, the mismatch travels upward as a prediction error, and either the guess gets revised or the body acts to make the world fit the forecast.
A feature called precision-weighting governs which signals get listened to. In a fog, the brain trusts touch over sight. In an unfamiliar city, it weights the map heavily and discounts the gut sense of direction. Emotions, on Bae’s account, are products of this same architecture, not a separate emotional system bolted onto a cognitive one but the same forecaster running predictions about what your body and the world together mean for you.
The Concepts That Build the Feeling
The active ingredient is what Bae calls an emotion concept, building on the philosopher Christian Michel and the neuroscientist Lisa Feldman Barrett. These aren’t dictionary entries. They’re “structured, hierarchical priors,” sprawling trees of expectations about how a particular kind of feeling tends to unfold. The concept FEAR knots together expectations about the body (quickened pulse, tensed shoulders), the world (something dangerous nearby), and the self (an urge to retreat, a narrowing of attention). When the brain categorizes a current swirl of signals under this concept, those signals stop being noise and become an emotion.
Emotions arise “when interoceptive and exteroceptive inputs are categorized under emotion concepts” inside the predictive model. The concept isn’t a label slapped onto a finished feeling. It’s part of what makes the feeling that feeling. Anger and fear can ride atop the very same arousal because the concepts pulling them together draw on different contexts, memories, and expectations.
The hiker’s snake makes the cleanest example. Heart rate climbs, attention narrows, the body leans away. Within the predictive hierarchy, the brain infers that the situation instantiates DANGER and binds the bodily preparation under the FEAR concept. Conscious fear, on this picture, is not read off the body. It’s the result of multiple streams of prediction being braided together under a single conceptual umbrella, an act of binding that gives the moment its specific shape.
Why Two People Never Feel the Same Anger
Because emotion concepts are learned, they vary. One person’s shame is anchored in social withdrawal, the urge to disappear; another’s involves anger or exposure, the urge to lash back. Anger in one culture means confrontation; in another, dignified silence. None of this is mysterious if emotions are constructions, because the construction kit differs from one childhood to another. Even fear itself can show up as “panicked escape in one context and quiet dread in another.”
The framework also makes room for emotions in infants and other animals, who lack adult-grade concepts. Bae proposes they operate with rudimentary generative models structured around basic goals like safety and familiarity, which sharpen over time into the more differentiated emotions we know. A baby’s distress is initially undifferentiated; it only gradually splinters into fear, frustration, and grief as the conceptual repertoire grows.
It also makes sense of the strange cases. The soldier who feels fear in battle and yet advances. The driver who only registers terror after the near-collision is over. The same arousal is being routed through different concepts and different forecasts about what the situation demands. Emotions, on this account, aren’t pushed up from below by a stricken body. They’re shaped from above by what the brain has, over a lifetime, learned to expect.
Why This Matters
If feelings are forecasts, then chronic emotional suffering can be read as forecasting gone wrong. Bae notes that anxiety can be modeled as the brain placing too much confidence in threat predictions, so that even disconfirming evidence fails to soften them. Depression looks like a system whose priors have grown reliably bleak, biasing inference toward despair. These framings, drawing on work by Anil Seth and others, suggest that therapy is partly the slow business of teaching the brain to make different bets.
The picture also reshapes how we might think about self-knowledge. If emotions are constructed, asking what we’re “really” feeling has no hidden answer waiting to be decoded from the body. The feeling is the construction, and learning new emotion concepts, through reading, conversation, or therapy, can genuinely change what becomes available to feel. A wider emotional vocabulary isn’t decoration; it’s a wider set of priors the brain can use.
Bae’s account doesn’t dissolve the strangeness of feeling. A pounding heart still pounds; grief still rearranges a familiar kitchen into a foreign room. What changes is the picture of where the meaning comes from. Not from a snake or a loss alone, and not from the body alone, but from the brain’s restless work of guessing what these signals add up to.