The Cognitive Scientists AI Forgot to Thank

There is a kind of forgetting that is not accidental. It is the forgetting of the person who climbs the ladder and, once on the roof, kicks it away and insists they simply grew wings. I know that move. I grew up inside a theology that had a genius for it: a system that took everything good in a person, relabeled it as a gift, and then charged rent. You did not build your character. You did not earn your kindness. It was borrowed, and the debt was eternal.

So forgive me if I recognize the pattern when I see a trillion-dollar industry standing on a century of public science and speaking as though it emerged fully formed from a venture-backed cloud.

I have a confession to make, and it is not a comfortable one. I am writing this with an AI assistant. I am using the same kind of tool built by the same industry I am critiquing. I cannot pretend to stand outside this system and analyze it from a place of purity. None of us can. We participate in structures we wish were different. We benefit from systems we recognize as broken. That is the shape of living inside the contradiction: you need the ladder to climb, even as you watch someone pull it up behind them.

I am not interested in performing innocence. I am interested in clarity. So let me be clear.

Modern AI did not appear out of Silicon Valley genius alone. It rests on decades of publicly funded work in neuroscience, psychology, cognitive science, statistics, and computer science — much of it done long before anyone knew how to monetize intelligence at scale. The companies profiting most extravagantly from that inheritance now speak as though the inheritance were irrelevant. That omission is not just historically careless. It is morally revealing.

I. The Question Came Before the Product

The earliest artificial neural networks were not built to sell software. They were attempts to answer an older question: how does thought arise from matter?

In 1943, Warren McCulloch and Walter Pitts published a paper on logical models of nervous activity. The title itself tells you what kind of project this was. Not a product roadmap. Not a startup pitch. A theory of mind. They were trying to understand what it means for physical tissue to produce cognition.

Then came Donald Hebb, who gave us a learning principle so durable it became common language: cells that fire together strengthen their connection. Again, this was not an optimization trick in service of ad monetization. It was an attempt to explain how learning happens in living systems.

And then Frank Rosenblatt, writing in a psychology journal, proposed the perceptron — another model born from the effort to understand information storage and organization in brains. These were not founders in the modern sense. They were researchers with questions.

That distinction matters.

Because one of the great myths of our era is that the most important technologies arrive from pure entrepreneurial brilliance. But in this case, the intellectual ancestors of AI were not raised in demo-day culture. They were raised in laboratories, departments, and grant-funded institutions that assumed the world should try to understand itself before it tried to extract rent from the result.

II. What Cats Taught Machines

The story becomes more concrete in vision science.

In the 1960s, David Hubel and Torsten Wiesel studied the cat visual cortex and showed that visual information is processed in layers: simple features first, then more complex structures, then abstraction. That work won a Nobel Prize, and it changed the way scientists thought about perception.

Decades later, that lineage helped inspire convolutional neural networks — the architecture that underlies much of modern computer vision. The point is not that a cat invented image recognition. The point is that a public, curiosity-driven investigation into how brains process images became part of the conceptual scaffolding for machine perception.

That is the pattern again: observation before application, theory before product.

And it is not just vision. In reinforcement learning, a crucial idea is the gap between what is expected and what is received. Wolfram Schultz’s work on reward prediction error helped sharpen the understanding that dopamine neurons encode not merely reward, but surprise relative to expectation. That insight became deeply relevant to machine learning, even if the engineering implementation was later and distinct.

This is the part people flatten into triumphal mythology: they say the industry invented everything it uses. It did not. It assembled, refined, and scaled ideas that were already being explored for reasons far older and more human than profit.

III. The Long Winter

There is another thing the mythology omits: this field did not march forward on an unbroken runway of obvious success. It nearly died, more than once.

AI winters are not just a footnote. They are a warning. When the hype cycle outruns the science, funding retreats, institutions lose patience, and the people doing the hardest work are forced to keep going on thin support and stubborn conviction. The Lighthill report in the 1970s helped trigger one of those winters by declaring the field a dead end. The collapse of the Fifth Generation project did the same in another era. The field survived because universities, public agencies, and research labs kept paying people to ask questions that did not yet have commercial answers.

That includes the long arc of backpropagation and connectionist research, much of it developed in academic settings by people trying to model cognition and learning, not build a marketable assistant. It includes decades of work in psychology, statistics, control theory, optimization, and computational neuroscience. It includes the kind of slow knowledge that does not fit neatly into quarterly reporting.

This is what a real intellectual inheritance looks like. It is messy, plural, cumulative, and often underappreciated by the generation that arrives after the hard part is done.

IV. The Debt

Here is the irony I cannot ignore.

The institutions that made this field possible — the public agencies, universities, and grant systems that paid for the basic science — are the very institutions that are so often treated as disposable once the private sector has learned enough to capitalize the results. That is the real theft of the story. Not the theft of code, but the theft of memory.

Because once the product works, the origin story gets rewritten. It becomes a fable of exceptional founders, visionary companies, and inevitable progress. The ladder disappears. The public disappears. The decades disappear. And the people who did the unglamorous work of asking how minds think, how neurons learn, how attention works, how reward is encoded, are rendered invisible in the market’s preferred narrative of self-creation.

I do not think that is just bad history. I think it is a civic failure.

A society that defunds basic research while celebrating the commercial harvest of basic research is not practicing prudence. It is eating the seed corn and calling it efficiency. It is benefiting from long horizons while refusing to support long horizons. It is demanding miracles on a spreadsheet and then sneering at the institutions that make miracles thinkable in the first place.

So yes, I am grateful to the engineers who built these systems. I am also unwilling to pretend they built them alone.

Modern AI is not the triumph of a lone industry inventing intelligence ex nihilo. It is the downstream product of generations of neuroscientists, psychologists, cognitive scientists, mathematicians, and computer scientists — many of them public servants in all but name — who asked, patiently and without guarantee, how does a mind work?

The honest answer to that question is not “we did this by ourselves.”

The honest answer is: we inherited it.

Footnotes

[1] McCulloch, W.S. and Pitts, W. “A Logical Calculus of the Ideas Immanent in Nervous Activity.” Bulletin of Mathematical Biophysics, Vol. 5, pp. 115-133, 1943.
https://doi.org/10.1007/BF02478259

[2] Hebb, D.O. The Organization of Behavior: A Neuropsychological Theory. Wiley, 1949.
https://books.google.com/books/about/The_Organization_of_Behavior.html?id=4F9uAAAAMAAJ

[3] Rosenblatt, F. “The Perceptron: A Probabilistic Model for Information Storage and Organization in the Brain.” Psychological Review, Vol. 65, No. 6, pp. 386-408, 1958.
https://doi.org/10.1037/h0042519

[4] Hubel, D.H. and Wiesel, T.N. “Receptive Fields, Binocular Interaction and Functional Architecture in the Cat’s Visual Cortex.” The Journal of Physiology, Vol. 160, pp. 106-154, 1962.
https://doi.org/10.1113/jphysiol.1962.sp006837

[5] Fukushima, K. “Neocognitron: A Self-organizing Neural Network Model for a Mechanism of Pattern Recognition Unaffected by Shift in Position.” Biological Cybernetics, Vol. 36, pp. 193-202, 1980.
https://doi.org/10.1007/BF00344251

[6] Zador, A.M. et al. “Catalyzing Next-Generation Artificial Intelligence through NeuroAI.” Nature Communications, Vol. 14, 1597, 2023.
https://doi.org/10.1038/s41467-023-37180-x

[7] Schultz, W., Dayan, P., and Montague, P.R. “A Neural Substrate of Prediction and Reward.” Science, Vol. 275, No. 5306, pp. 1593-1599, 1997.
https://doi.org/10.1126/science.275.5306.1593

[8] Lighthill, J. “Artificial Intelligence: A General Survey.” Science Research Council, 1973.
https://www.aiai.ed.ac.uk/events/lighthill1973/lighthill.pdf

[9] Rumelhart, D.E., Hinton, G.E., and Williams, R.J. “Learning Representations by Back-Propagating Errors.” Nature, Vol. 323, pp. 533-536, 1986.
https://doi.org/10.1038/323533a0

[10] Rumelhart, D.E. and McClelland, J.L. Parallel Distributed Processing. MIT Press, 1986.
https://en.wikipedia.org/wiki/Parallel_Distributed_Processing

[11] Bahdanau, D., Cho, K., and Bengio, Y. “Neural Machine Translation by Jointly Learning to Align and Translate.” arXiv:1409.0473, 2014.
https://arxiv.org/abs/1409.0473

[12] Vaswani, A. et al. “Attention Is All You Need.” NeurIPS, 2017.
https://arxiv.org/abs/1706.03762

[13] Treisman, A.M. and Gelade, G. “A Feature-Integration Theory of Attention.” Cognitive Psychology, Vol. 12, No. 1, pp. 97-136, 1980.
https://doi.org/10.1016/0010-0285(80)90005-5

[14] Posner, M.I. “Orienting of Attention.” Quarterly Journal of Experimental Psychology, Vol. 32, No. 1, pp. 3-25, 1980.
https://doi.org/10.1080/00335558008248231

[15] Rescorla, R.A. and Wagner, A.R. “A Theory of Pavlovian Conditioning.” In Classical Conditioning II, pp. 64-99, 1972.
http://www.scholarpedia.org/article/Rescorla-Wagner_model