Why science cannot prove or disprove free will

Leftist-atheist-evolutionist says that he was persuaded that he has no free will by this PNAS article:

The beauty of the mind of man has nothing to do with free will or any unique hold that biology has on select laws of physics or chemistry. This beauty lies in the complexity of the chemistry and cell biology of the brain, which enables a select few of us to compose like Mozart and Verdi, and the rest of us to appreciate listening to these compositions. The reality is, not only do we have no more free will than a fly or a bacterium, in actuality we have no more free will than a bowl of sugar. The laws of nature are uniform throughout, and these laws do not accommodate the concept of free will.

I mention Coyne's views here. That article yielded this rebuttal:

In his reply (1) to the letter by Anckarsäter (2) commenting on his original article (3), Anthony Cashmore expresses the view that a belief in free will would require at least a molecular model as a justification. However, such a model cannot exist, as I will explain in the following.

The behavior of an agent possessing free will is by definition unpredictable. In contrast to stochastic phenomena, it is not even possible to predict all observable statistical properties of the behavior of such an agent. A molecular model for free will, or in fact any scientific model for free will, would thus have to contain some property labeled as unpredictable.

However, the scientific method that we apply today, which is based on the formulation of hypotheses that are then tested by observation and experiment, cannot accommodate unpredictability. The statement that "property X is unpredictable" cannot be tested by observation and is thus not a scientific hypothesis. Moreover, even if property X itself is observable, its supposed unpredictability makes it impossible to formulate scientific hypotheses about it. As a consequence, free will cannot be integrated into any scientific model.

The only way in which the scientific method could resolve the question of the existence of free will is by showing its nonexistence. This would require a scientific model that permits a complete prediction of human behavior, or at least of all its observable statistical properties. However, as Anckarsäter pointed out (2), we are very far from having such a model.

Cashmore goes on to claim that in the absence of a good reason to believe in free will, we should believe in its nonexistence. A pragmatically minded person would counter that, in the absence of solid evidence to the contrary, we should trust our perception, which tells us that we do have free will. However, neither point of view can claim science as its justification. For a believer in the scientific method, the only coherent point of view is agnosticism about the existence of free will.

I agree that free will cannot be scientifically proved or disproved.

I also agree that our perception gives us good reason to believe in free will, even if it cannot be proved.

He says: "A molecular model for free will ... would thus have to contain some property labeled as unpredictable." Yes, that is right, and our best molecular models do indeed have properties labeled as unpredictable. All quantum systems do. Whether that unpredictability has anything to do with free will is an open question. The question may never be solved, for the reasons he gives. Maybe molecules have a tiny bit of free will, and human free will is derived from the collective free will of molecules.