Skip to content

2026 08 17 Why Won't Computer Functionalism Die? Gemini Chat

Despite decades of blistering attacks from philosophers like John Searle, Hubert Dreyfus, and Hilary Putnam, as well as neuroscientists and psychologists like Robert Epstein, computer functionalism refuses to die.

It persists not because its critics are wrong, but because functionalism is methodologically indispensable, mathematically tractable, and ideologically comfortable. In science, a flawed theory is rarely abandoned simply because it has philosophical holes; it is only abandoned when a more productive, practical alternative replaces it.

1. The Methodological Monopoly (It Gives Scientists Something to Measure)

The primary reason computational functionalism survives is practical: it works as an experimental research program.

  • Measurable Variables: If the brain is an information-processing system, researchers can measure "processing speed," "working memory capacity," "signal-to-noise ratios," "attentional bandwidth," and "error rates."

  • Operational Hypotheses: Cognitive psychologists can design clean, falsifiable lab experiments testing how inputs transform into outputs.

  • The Alternative Problem: If you tell a cognitive scientist that the brain doesn't process information, but is instead an irreducible, dynamic biological swirl of hormones, history, and bodily coping (as Dreyfus or Epstein suggest), the immediate lab response is: "Okay... so what equation do I run on Monday morning?"

2. High-Level Cognition Resists Purely Embodied Models

Alternatives like 4E Cognition and Dynamical Systems Theory (DST) excel at explaining sensorimotor tasks—such as an outfielder catching a fly ball, an infant learning to walk, or an animal navigating terrain.

However, these frameworks struggle when trying to explain abstract, disembodied reasoning:

           Task Domain                       Best-Fitting Model
┌──────────────────────────────────────┐  ┌──────────────────────────────────────┐
│ Sensorimotor Coordination            │  │ 4E Cognition & Dynamical Systems     │
│ (Walking, catching, tactile balance) │  │ (Feedback loops, continuous physics) │
├──────────────────────────────────────┤  ├──────────────────────────────────────┤
│ Abstract Symbolic Reasoning          │  │ Computational / Functional Models    │
│ (Playing chess, filing taxes, logic) │  │ (State transitions, representations)  │
└──────────────────────────────────────┘  └──────────────────────────────────────┘

When a human plans a five-year retirement budget, proves a geometric theorem, or debugs an institutional SQL query, they are manipulating abstract counterfactuals and symbolic structures detached from immediate sensory inputs. Computational functionalism remains the only framework with an established mathematical language (Turing computation, Boolean logic, Bayesian inference) capable of modeling these high-level cognitive tasks.

3. The Connectionist Pivot (Functionalism Mutated)

When classical symbolic AI (GOFAI) hit a wall in the 1980s and 90s, critics declared the death of computationalism. But functionalism didn't die—it simply mutated into Connectionism and Neural Networks.

  • Sub-symbolic Functionalism: Researchers abandoned rigid if/then algorithms in favor of distributed, weighted matrices and probabilistic activation functions.

  • Empirical Success: Modern deep learning and large language models proved that complex, human-like linguistic and reasoning capabilities can emerge from matrix multiplication and next-token prediction.

  • This practical success breathed massive new life into functionalism: even if biological brains don't work like von Neumann desktop PCs, they still appear to perform complex mathematical optimization across distributed nodes.

4. The Fear of Dualism and "Carbon Chauvinism"

In analytic philosophy, functionalism remains the most popular way to defend materialism without collapsing into reductionism:

  • Avoiding Cartesian Dualism: It explains mental life using purely physical causality without invoking non-physical souls or mysterious substances.

  • Avoiding Mind-Brain Identity Chauvinism: If mental states are strictly identical to human biological neurons (as Searle’s Biological Naturalism suggests), then an intelligent alien with silicon-based biology, or an octopus with a radically different nervous system, could never technically have a "belief" or "pain." Functionalism’s doctrine of multiple realizability feels intuitively broad and scientifically egalitarian.

When critics reject functionalism, physicalists worry the only alternatives are mysticism (treating biological tissue as having magical, non-computable properties) or eliminativism (declaring that thoughts and feelings don't exist at all).

5. Economic and Cultural Momentum

Finally, the metaphor survives because of sheer institutional gravity:

"We shape our tools, and thereafter our tools shape us." — Often attributed to Marshall McLuhan / John Culkin

  • The Tech Industry: Trillions of dollars, massive research grants, and entire university departments are organized around computer science, machine learning, and computational neuroscience.

  • Linguistic Saturation: The computational metaphor has colonized our everyday language. People routinely say they "lack the bandwidth" for a task, need to "process" bad news, or are "hardwired" for certain behaviors. Once a metaphor becomes the vernacular of everyday life, root-level critiques struggle to gain mainstream traction.

Summary

Computer functionalism refuses to die because it is the only model that bridges the gap between mechanical causality and abstract thought.

Until cognitive science develops a formal mathematical framework that can explain high-level reasoning and language without invoking representations or information processing, functionalism will remain the default—even among scientists who privately acknowledge its philosophical flaws.