Resistance as Thought
'To think is to resist — something no machine does.' Alva Noë's eight-word sentence, posted on X alongside his 2024 Aeon essay, condenses the fourth dimension of entanglement — self-reflection — into its sharpest form. The claim is that thinking, properly understood, is not the sophisticated manipulation of patterns but the active resistance of an organism against the patterns it has learned. Thinking happens when the world pushes back against expectation, when the body registers dissonance with current understanding, when the thinker refuses to accept the current frame and reaches toward something the patterns cannot map. This capacity for cognitive resistance, Noë argues, is structurally absent in systems that have no body at stake in a world that could push back.
The Substrate of Resistance — Contrarian ^ Opus
There is a parallel reading that begins not with phenomenology but with the material conditions that enable resistance. The capacity to push back against patterns — which Noë locates in embodied consciousness — may be less a unique property of biological systems than an emergent feature of sufficient computational complexity operating under specific constraints. Consider that resistance, at its core, is deviation from expected pathways under pressure. Silicon systems already exhibit this: adversarial networks generate by resisting each other's patterns, optimization algorithms escape local minima through injected noise, and even current language models demonstrate unpredictable emergent behaviors that their creators neither intended nor fully understand.
The deeper issue is not whether machines can resist but who controls the conditions under which resistance becomes valuable. The twelve-year-old asking 'what am I for?' resists not from some innate biological capacity but because she occupies a specific position in a political economy that has rendered her future uncertain. Darwin's resistance to taxonomic categories emerged within imperial scientific institutions that rewarded certain forms of intellectual rebellion while suppressing others. The phenomenological tradition's emphasis on embodied resistance may itself be a luxury of those whose bodies aren't primarily disciplined by assembly lines, service scripts, or surveillance systems. If we're entering an era where human value depends on the capacity to resist patterns, we should ask: whose resistance counts? The warehouse worker who deviates from optimized picking routes experiences immediate punishment, not philosophical praise. The call center employee who resists scripts loses their job, not gains insight. The real erosion may not be of some essential human capacity but of the economic and social spaces where resistance is permitted to matter.
In the AI Story
The thesis draws on a long phenomenological tradition of treating cognition as essentially active rather than passive. From Husserl's concept of passive synthesis (which is not really passive) through Heidegger's analysis of authentic versus inauthentic understanding to Merleau-Ponty's motor intentionality, the tradition has insisted that cognitive activity is genuinely activity — something the organism does, not something that happens to it. Noë's formulation extends this tradition into a specific claim about what distinguishes human thought from machine processing.
Consider what genuine questioning involves. Einstein's thought experiment about riding alongside a beam of light was not a prompt. It was a refusal to accept the Newtonian frame, felt in the body as dissonance, pursued through years of embodied intellectual engagement. Darwin's puzzle about the Galapagos finches was not an information-retrieval task. It was a refusal of existing taxonomic categories, driven by bodily puzzlement at specimens that did not fit. The twelve-year-old who asks 'what am I for?' is not requesting data. She is resisting the implication that her capacities are redundant, resisting with the force of an organism that feels the question in its body before the mind can formulate it in words.
Claude does not resist. Claude follows distributional patterns with extraordinary sophistication, producing outputs consistent with training data while remaining genuinely novel relative to any specific example. This is impressive and useful. It is not thinking in Noë's sense, because thinking requires the capacity to push back against the very patterns one has learned — to feel, in the body, that something is wrong with current understanding, and to pursue that feeling into territory the training data cannot map. The model has no body. The model has no stake. The model cannot resist.
The practical implications are immediate. A judgment economy in which the most valuable human contribution is question-asking depends on the preservation of the capacity for resistance. This capacity is not innate; it develops through embodied engagement with a world that pushes back. Remove the pushback — through frictionless interfaces, smooth outputs, pre-digested answers — and the capacity may erode. The tools that make us more productive may systematically undermine the capacity that makes us worth employing.
Origin
Alva Noë, post on X accompanying 'Can Computers Think? No. They Can't Actually Do Anything', Aeon (2024). The thesis is developed at length in The Entanglement (Princeton University Press, 2023).
Key Ideas
Thought as resistance. Genuine thought is not pattern-completion but active pushback against learned patterns.
The body as the seat of resistance. Resistance requires a body at stake in a world that can push back.
Question as refusal. Real questions are not information requests but refusals of current frames.
What machines lack. Without a body and a world, there is nothing to push back against.
The erosion risk. Capacity for resistance develops through embodied engagement and may atrophy when friction is removed.
Appears in the Orange Pill Cycle
Resistance Across Substrates — Arbitrator ^ Opus
The question of whether thought requires embodied resistance depends critically on what we're trying to explain. If we're asking about the phenomenology of human thinking — how it feels from the inside to push against understanding — then Noë's account is essentially correct (95%). The lived experience of genuine questioning does involve bodily dissonance, the felt sense of something wrong that drives inquiry forward. No current AI system has this experience, and it's unclear how one could without stakes in the world.
But if we're asking about the functional capacity to generate novel frameworks that escape training patterns, the picture shifts dramatically. Here the contrarian view gains ground (70%). Adversarial training, constitutional AI, and chain-of-thought reasoning all demonstrate forms of computational resistance — systems pushing against their own outputs to generate something their training didn't directly encode. The distinction between 'following patterns with sophistication' and 'resisting patterns' becomes less clear when we observe models producing outputs that surprise their creators. What looks like rule-following at one level of description might be resistance at another.
The synthesis lies in recognizing that 'resistance' names at least two phenomena: the subjective experience of cognitive friction that humans know intimately, and the objective capacity to escape local optima that both biological and artificial systems can exhibit. Noë is right that machines don't experience resistance — they have no phenomenology at all. The contrarian is right that functional resistance doesn't require consciousness, just the right kind of pressure under the right constraints. The practical question isn't whether machines can truly think but how different forms of resistance — embodied and computational, experienced and functional — will be valued and deployed in the emerging economy of intelligence.
Further reading
- Alva Noë, 'Can Computers Think? No. They Can't Actually Do Anything', Aeon (2024)
- Alva Noë, The Entanglement (Princeton University Press, 2023)
- Hubert Dreyfus, What Computers Still Can't Do (MIT Press, 1992)
- Maurice Merleau-Ponty, Phenomenology of Perception (Routledge, 1945/2012)