We Feel, Therefore We Learn
The paper argued that the brain does not store information neutrally but encodes it in relation to its emotional significance — its relevance to the organism's goals, its connection to the learner's developing sense of self, its felt importance within the project of being alive. Learning that lacks emotional engagement produces a qualitatively different kind of knowledge: information stored but not integrated, accessible but not understood. The title inverts Descartes's cogito to signal the paper's ambition: to replace the model of cognition as dispassionate calculation with a framework in which feeling and thinking are dimensions of a single process, woven together at the level of neural architecture. The finding carries direct implications for AI-augmented education, where frictionless information delivery threatens to eliminate the emotional engagement that transforms information into understanding.
In the AI Story
The Enlightenment model of cognition — still dominating most educational institutions and most technology companies — holds that the best thinking is dispassionate thinking. Clear the mind of feeling. Suppress the body's signals. Attend only to the logical structure of the problem. Damasio's clinical work had already begun to dismantle this assumption through patients like Elliot, whose intact intellect and destroyed emotional processing produced not clearer thinking but no effective thinking at all.
Immordino-Yang's 2007 paper extended the dismantling from decision-making into learning. The argument ran: if emotion guides decision by attaching felt significance to cognitive representations, then learning — which is the acquisition of representations that will guide future decisions — requires emotional engagement to produce representations that carry the felt significance decisions will need.
The distinction between information and understanding is the hinge on which the entire relationship between human cognition and artificial intelligence turns. Claude can provide information with an efficiency no human teacher can match. The question Immordino-Yang's research forces is whether the student who receives this information has been educated — whether the information has been integrated in a way that will persist, transfer, and inform judgment.
The 2024 Royal Society theme issue on embodied cognition in the age of AI framed the stakes precisely: embodiment is not an incidental feature of human cognition but a constitutive one, and AI systems that process symbols without being in the world in an embodied sense operate on a fundamentally different basis than human minds.
Origin
The paper emerged from Immordino-Yang's doctoral work in Damasio's laboratory at USC's Brain and Creativity Institute, where the somatic marker hypothesis had been the central theoretical framework. The title's inversion of cogito ergo sum was deliberate provocation — a flag planted on territory that the educational establishment was not prepared to concede.
Key Ideas
Emotion and cognition are not separate processes. They are dimensions of a single process woven together at the level of neural architecture.
The brain encodes significance, not just content. Information is stored in relation to its emotional relevance, which determines whether it will be retrievable and actionable later.
Struggle is the mechanism of deep learning. The frustration-to-satisfaction arc tags experiences with significance that efficient delivery cannot replicate.
Education is development, not transfer. The purpose is the transformation of the learner, not the relocation of content — a distinction frictionless AI tutoring cannot honor.
If we do not feel, we do not learn; we merely accumulate. The corollary the productivity culture least wants to hear.
Debates & Critiques
Educational technology advocates argue that adaptive AI tutoring can produce emotional engagement through personalization, gamification, and responsive feedback. Immordino-Yang's rejoinder: these produce fast-emotion engagement (dopamine spikes correlated with achievement) rather than the slow, transcendent-emotion engagement that activates the deepest learning systems.
Appears in the Orange Pill Cycle
Further reading
- Mary Helen Immordino-Yang and Antonio Damasio, We Feel, Therefore We Learn: The Relevance of Affective and Social Neuroscience to Education (Mind, Brain, and Education, 2007)
- Antonio Damasio, Descartes' Error (Putnam, 1994)
- Mary Helen Immordino-Yang, Emotions, Learning, and the Brain (W.W. Norton, 2016)