Salamander

Here, I’m assuming “it” is a conscious perception. But now I’m confused again because I don’t think any theory of mind would deny this.

Yes, the example of such a theory that is common is epiphenomalism. What I am contrasting in my answers is the epiphenomalist/hard-determinist framework with the physicalist/compatibilist one.

stimuli -> CPM ⊆ brain -> consciousness update CPM -?> black box -?> mind -?> brain -> nervous system -> response to stimuli

I can try to explain with such a diagram:

stimuli -> nerves -> brain input ports -> brain filtering and distribution -> Conscious brain processing via causal predictive modelling -> brain output ports -> nerves -> conscious action
                                                                                          |
                                                                                          -- > Unconscious processing -> brain output ports -> nerves -> unconscious action

So, the CPM is a process within the brain. The idea is that the brain is a computer that makes predictions by building cause-and-effect models. What is interesting about the mathematics of causal models is that the underlying engine is the counterfactual. The claim being made here is that mind itself is this counterfactual engine doing its work. The computational space that deals with the counterfactuals or “fantasies” is the essence of the mind.

This is not in any way a solution to the hard problem of consciousness. Rather, it is one of many frameworks compatible with physicalism, and it is the one I personally subscribe to. In this framework, it is a postulate that conscious experience corresponds to the brain’s counterfactual simulations within a generative model used for predicting and guiding action. This postulate does not prove or mechanistically explain consciousness. No physical theory currently does.

I’m going to stick with the meat of your point. To summarize, …

That is not quite how I see it. The linear diagram “brain -> black box -> mind” represents a common mode of thinking about the mind as a by-product of complex brain activity. Modern theories are a lot more integrative. Conscious perception is not just a byproduct of the form brain -> black box -> mind, but instead it is an essential active element in the thought process.

Ascribing predictions, fantasies, and hypotheses to the brain or calling it a statistical organ sidesteps the hard problem and collapses it into a physicalist view. They don’t posit a mind-body relationship, they speak about body and never acknowledge the mind. I find this frustrating.

That text was probably written by a materialist / physicalist, and this view is consistent within this framework. It is OK that you find this frustrating, and it is also alright if you don’t accept the materialist / physicalist viewpoint. I am not making an argument about materialism being the ultimate truth, or about materialism having all of the answers - especially not answers relating to the hard problem! I am specifically describing how different frameworks held by people who already hold a materialist view can lead to different ways of understanding free will.

Scientists often do sidestep the hard problem in the sense that they acknowledge it to be “hard” and keep moving without dwelling on it. There are many philosophers (David Chalmers, Daniel Dennett, Stuart R. Hameroff), that do like getting into the nitty-gritty about the hard problem, so there is plenty of material about it, but the general consensus is that the answers to the hard problem cannot be find using the materialist’s toolkit.

Materialists have is a mechanism for building consensus via the scientific method. This consensus mechanism has allowed us to understand a lot about the world. I share your frustration in that this class of methods does not seem to be capable of solving the hard problem.

We may never discover a mechanism to build consensus on the hard problem, and unfortunately this means that answers to many very important questions will remain subjective. As an example, if we eventually implement active inference into a computer, and the computer claims to be conscious, we may have no consensus mechanism to determine whether they “really” are conscious or not, just as we cannot ascertain today whether the people around us are conscious. In my opinion, yes, it is physically possible to build conscious systems, and at some point it will get tricky because it will remain a matter of opinion. It will be an extremely polarizing topic.

In my view, neuroscience may contribute to clarifying questions like:

• Do all brains support a conscious predictive model (CPM)?

• Does adaptive behavior in brainless organisms suggest a primitive CPM?

• What is the relationship between brain and mind?

But deeper questions, such as “What do we mean by mind?” or “Why assume weak emergence?” remain tied to the hard problem of consciousness, which currently lies beyond the reach of empirical science.

In trying to describe promising cognitive models, I buried my main point. I am not arguing that the brain and mind problem is close to a solution, or that science is close to resolving it.

Here is my actual point:

Certain materialist views unintentionally reproduce dualist thinking. Substance dualism claims that the mind exists outside physical law. Materialism, in contrast, holds that the mind emerges from brain activity. But when this emergence is explained only as complexity or undefined processing, a conceptual gap forms: brain -> black box -> mind. This reproduces dualism in practice, even if not in theory.

This gap renders consciousness a passive byproduct. It becomes a new kind of soul, unable to influence the body. A mind without agency.

Predictive processing and active inference models offer an alternative. They describe the brain as a generative system that continuously updates predictions based on sensory input. As summarized in a recent review:

Active inference casts the brain as a fantastic organ: a generator of fantasies, hypotheses and predictions that are tested against sensory evidence.

While these models do not resolve the hard problem, they help remove part of the black box. They suggest that consciousness may play a functional role in these feedback loops. It is not a detached illusion but a process embedded in how the brain operates.

For me, this shift changed how I think about free will. Not because it provides final answers, but because it allows me to see mental acts in a similar way to how I see muscle movement. These acts are constrained by physical laws, but they are still mine.

I think that its emergence is weak but I see no resolution to the hard problem of consciousness any time soon, so for the time my opinions about it are ideas that I find compelling and intuitive and not grounded in facts and evidence. Weak emergence does posit some form of pansychism in the sense that sentient-like behavior can emerge in other brains and even that characteristics that we might associate with sentience might emerge from other phenomena present through the universe. But, because of the same reasons that the hard problem is hard, it is also hard to study and learn about these phenomena.

I can try to explain a little better what I meant.

I don’t believe we have “free will” in the sense that the mind is separate from the body (dualism) and that it is able to break the laws of physics by altering our physiological processes. I don’t think that the non-determinism of quantum mechanics in itself gives us agency, and our mind does not have a mechanism to select how a particular wavev function collapses (not a fan of the Orch OR model).

So, in this traditional sense my answer is “no, we do not have free will”

But I think that the existential crisis and feeling of a lack of agency stems from the model of sentience that one believes. If one rejects dualism, posits that consciousness is an emergent property of the brain, but then ascribes only very loosely a mechanism to consciousness such as ‘complex information processing gives rise to consciousness’, then sentience appears to be just some unexplained quirk that is not essential and just happens to be there. Combining a lack of dualism and free will with consciousness being a useless quirk is what (I think) creates the existential crisis associated with a lack of free will. I used to fall into this camp of thought and resolved the crisis through a logic such as: “Yeah, there is no free will, living is nice though so I am happy that I can accidentally experience the world”.

What pushed me to re-assess this way of thinking originally was reading through a paper about teaching a dish of neurons how to play pong](https://www.cell.com/neuron/fulltext/S0896-6273(22)00806-6). At first it did not make sense to me how one can possibly provide feedback to a group of isolated neurons such that it could learn to play a game. What ‘reward’ can you give a group of neurons to push them to do what you want?!

I looked into Karl Friston, the last author of that paper, which led me down a path of study. I discovered Judea Pearl, who formalized causal reasoning in a way that lets us build statistical models to move from correlations to counterfactual causes. This makes it possible to teach causal inference even to machines.

Karl Friston’s work and other researchers in the field argue that the brain is a computer built for causal computing. This idea underpins the Bayesian brain, Predictive Coding Theory, Active Inference.

In Karl Friston’s Active Inference book, sentience is proposed to emerge as a result of the prediction engine. What we experience is not actually what our senses already experienced, but instead it is what our brain expects that we will sense in the next instant. This model of reality that is built by our brain in its attempt to perform its basic function (link causes to effects in order to predict the next stimulus).

One idea is that consciousness emerges because the predictive brain is creating a ‘model’ that does not exist in physical space and so it needs imagination to explore it. The imagination of things that do not exist is essential to the process of generating counterfactuals, and counterfactuals are at the core of the causality machine. To show that A causes B, you need to imagine a situation in which A is not present and estimate the likelyhood of B. One idea is that it is precisely in the creation of a world without A that sentience emerges.

A lot of these ideas are not falsifiable, so it is difficult to say that this is indeed the mechanism of consciousness. But some of the ideas are falsifiable, and those ideas have helped these researchers teach neurons how to play pong, so I think they might have a point.

So, then, I find it plausible that consciousness is not a quirk but an essential feature of our brain. To me this resolves the free will crisis because my consciousness is not an accidental outcome of physical processes just chaotically whizzing by but an actual feature of the machinery that is me.

So these outputs are experienced. And that experience of making predictions is me. Am I the one experiencing the predictions as well?

So this sentence confuses me: “This prediction machine is me making predictions and choices.” Am I making the predictions or is it the CPM?

I am this machine and I follow the laws of physics. I am part of physical reality, and my sentience is a feature of who I am. If I do something it is because I chose to do so, and the fact that I chose to do so in accordance to the law of physics does not remove my agency.

Thoughts and muscle movements come about through the opening and closing of ion channels that allow information to travel through neurons and for muscle fibers to contract and relax. ‘Free will’ in the sense that our mind is separate from our body and that it can somehow open those ion channels is a combination of dualism and molecular telekinesis, so I do not believe that, no.

But I do believe that consciousness is an essential emergent property of our brain. What we experience might be the output of a causal prediction engine in our brain that is making a prediction about the immediate sensory experience in a way that we can respond to stimuli before they happen. In that sense, yes, I do believe in free will because that conscious output that I experience is me! This prediction machine is me making predictions and choices.

I think that a materialist framing of free will requires accepting some model of consciousness in which consciousness is not just a weird accident but is a physical phenomenon that is part of us. An essential feature of how our brain works. This is not yet demonstrated (very difficult if not impossible to do so), but I think it is. Then ‘free will’ and ‘a material system following the laws of physics’ is no longer a contradiction.