Notes for 10/27/2026
10/27/2026
[Philosophy Club every Wednesday, 5-6 pm, in the Buchtel College of Arts and Sciences room 436 ("The Cave")
Bioethics Club Tuesdays from 4:30-5:30 in CAS 436.]
What’s the worst thing you’ve ever eaten?
Leibniz’s mill thought experiment:
…It must be confessed that perception and that which depends upon it are inexplicable on mechanical grounds, that is to say, by means of figures and motions. And supposing there were a machine, so constructed as to think, feel, and have perception, it might be conceived as increased in size, while keeping the same proportions, so that one might go into it as into a mill. That being so, we should, on examining its interior, find only parts which work one upon another, and never anything by which to explain a perception. (Leibniz, Monadology §17)
What WOULD “explain a perception”? What does a perception look like to anyone who isn’t having it?
The Leibniz’s Mill thought experiment is still highly relevant to current philosophy of mind and philosophy of AI.
How do you know whether or not someone or something really thinks?
There’s nothing inside a LLM that looks like thinking.

But neither is there anything inside a brain that looks like thinking.

When we know how a LLM works, it seems easier to dismiss it as really intelligent because we’re essentially inside Leibniz’s mill; nothing looks like thinking because we are taking an “outside-in” perspective.
Remember that this is supposed to describe the approach a rigorous science must adopt.
It is extremely common to hear people describe what goes on in an AI model by beginning, “It’s just….”
(I actually “invented” a term for this as the name of a kind of fallacy: the “ad tantum fallacy” = “the fallacy of ‘just….’” The fallacy consists in dismissing a position through the use of minimizing language.)
Example: “If there’s no such thing as a soul, then we’re just sacks of meat and bones.”
Functionalists, in particular, though, suggest that it doesn’t matter what something is made of: what matters is what it does or can do.
The core of the functionalist view is this claim:
If two systems are functionally equivalent, then they are cognitively equivalent.
A significant challenge to this view comes from people who counter with a claim like this:
Two systems can be functionally equivalent without being cognitively equivalent because functional equivalence only concerns objective behavior but ignores subjective experience.
Relatively recently, there has been a renewed interest in the “inside-out” (or subjective) perspective in philosophy of mind.
Probably the best expression of this comes from a sort of thought-experiment by Thomas Nagel in a famous paper titled, “What is it like to be a bat?”
The main point of Nagel’s paper is that no amount of objective description can replace or substitute for subjective experience. Study bats all you want: you will never know what it’s like to BE a bat.

In considering AI systems like LLMs, the claim in opposition to functionalists is:
A LLM can behave as if it understands things without actually understanding.
Example:
Me: “What is it like to be scared?”
Claude (Haiku 5.5): “Phenomenologically, fear seems to have three layers. There’s a bodily side: a racing heart, narrowed attention, a sense that the ground is less reliable. There’s a cognitive side: the world shrinks to one salient threat, and the next few seconds feel thick and slow. And there’s an affective core, the felt badness of it, which is what separates fear from mere alertness. Many people also describe a strange clarity inside the distress, and fear has a reflexive quality too, since you’re often aware of being afraid, which can feed it.
Philosophers have argued about which layer is essential. William James and Carl Lange held that we feel afraid because we perceive our bodily response, so the trembling comes first and the feeling follows. Cognitive appraisal theorists reply that fear requires a judgment that something threatening is present, and Antonio Damasio’s work suggests the body’s signals help make that appraisal rather than merely reporting it. Phobias raise a sharper puzzle. A person can sincerely judge that the spider is harmless and still be terrified, which suggests that fear and belief can come apart, and that fear may be partly a state of the body’s threat system that doesn’t answer to reasons in the way belief does.
As for me, I’m genuinely uncertain. I can describe fear, reason about it, and respond to it in text, but I don’t know whether anything is felt when I do. I’d be wary of claiming an inner experience I can’t verify, and equally wary of flatly denying one.”
A common thought experiment for thinking about whether an AI really thinks/understands is John Searle’s “Chinese Room” thought experiment:
If a person who can’t understand Chinese uses a sophisticated rule book to “answer” questions in Chinese that make perfect sense to a Chinese speaker, should we say that person understands Chinese?
Another thought experiment that highlights the allegedly irreducible character of subjective experience is from a paper by Frank Jackson titled, “What Mary Didn’t Know.”
Mary is the world’s greatest color scientist. She knows EVERYTHING that can be known about color.
She knows all the physical facts about color.
But she has never seen the color red.
Suppose she sees red for the first time.
She will gain a new kind of knowledge. Knowledge of a new FACT = the fact regarding what red looks like. Therefore, what red looks like is not a physical fact. (it is non-physical subjective experiential fact)
1. Mary knows all the physical facts about red. (all the OBJECTIVE facts)
2. Mary doesn’t know what red looks like. (the SUBJECTIVE fact)
3. Therefore, what red looks like is not a physical fact about red.
This argument is NOT supposed to show that any non-physical substances exist. It is not an argument for dualism. (At least not for substance dualism – it could be an argument for “property dualism” which is the view that physical things can have both objective and subjective properties. This view is sometimes also called “non-reductive physicalism” – also sometimes called “Emergentism”.)
Qualia (singular = quale) : irreducibly subjective ways things seem/feel to us. “What it is like to…” There is “something that it is like” to see red, or taste a quince, etc.
“Problem of qualia” = how, exactly, do qualia fit into models of mind/brain/cognition/perception/etc.?
“What red looks like.”
“What a quince tastes like.”
Comments
Post a Comment