Determinism and free will

[JimC]

Predictive text is a rather simple algorithm. If predictive text has "free will" then any conditional programming statement has free will. If-Elseif-Elseif-Else would qualify.

That's what rEv keeps saying: free will, in the sense of a more or less autonomous agency, is an absurd concept.

Free will, as something "extra" added to human biology, is indeed absurd, as absurd as the concept of a soul...

Dennett's idea is nothing to do with some added autonomous ability, but simply a way to escape the fatalism inherent in the robotic, Lagrangian view of humans as fully and utterly determined entities, where the concept of choice is meaningless.

I think that "choice" is a perfectly reasonable description of the way in which humans, evolved as very complex agents indeed, negotiate their way through the tricky territory of life abounding with other clever agents, whose best interests are not always yours. Of course much our our choices are constrained and affected by the past, but given both the complexity of these factors, and (IMO) some chaos inherent in our neurobiology, there is no way that choices could ever be fully predicted, even by a Lagrangian super being.

It's the relative aspect which I find most convincing. People born with severe intellectual disabilities have less effective agency, fewer choices and less of Dennett's free will. Even for humans with no biological disability, aspects of their political environment can, to an extent, constrain or extend their personal degrees of freedom.

[pErvinalia]

I've suggested the predictive text analogy doesn't work as a response to what Scot said, because it doesnt account for the common notion of free will as motive origination, and invited you to expand on your thinking. That invitation remains open.

I replied to that already. Scott didn't say anything about "motive origination". Whatever that is.

[Brian Peacock]

I've suggested the predictive text analogy doesn't work as a response to what Scot said, because it doesnt account for the common notion of free will as motive origination, and invited you to expand on your thinking. That invitation remains open.

I replied to that already. Scott didn't say anything about "motive origination". Whatever that is.

'Origination' is a term similar in scope to 'cause', that in which the state of a system originates, where motive is, well, motives: thus it relates to specific kinds of causes, or agency - to contribute to the cause of a set of circumstance through active agency, is an even more long-winded way of putting it. I suggested that the mere existence of alternatives does not comprise free will and that the predictive txt algorithm did not qualify as agency, for the reasons given. I'm interested in your views on this and the thinking behind them here, and so the aforementioned invitation remains open.

[JimC]

I've suggested the predictive text analogy doesn't work as a response to what Scot said, because it doesnt account for the common notion of free will as motive origination, and invited you to expand on your thinking. That invitation remains open.

I replied to that already. Scott didn't say anything about "motive origination". Whatever that is.

'Origination' is a term similar in scope to 'cause', that in which the state of a system originates, where motive is, well, motives: thus it relates to specific kinds of causes, or agency - to contribute to the cause of a set of circumstance through active agency, is an even more long-winded way of putting it. I suggested that the mere existence of alternatives does not comprise free will and that the predictive txt algorithm did not qualify as agency, for the reasons given. I'm interested in your views on this and the thinking behind them here, and so the aforementioned invitation remains open.

Dennett would say that agency is something that exists in degrees. It would be fair enough to say that a rock has zero agency, but the moment that either a designed or evolved entity has even a simple form of responding to an environmental condition (if x, do y), then we have the faint beginning of agency. Both bacteria and a variety of simple programs would seem to fit the bill to be simple agents. This can be a fully determined agency (if x, then always y), but could also have a stochastic element (if x, then 30% of the time do z, and 70% do y), which would seem at least to involve a somewhat wider range of freedom.

As for motives, perhaps they can be seen as reasons for action, reasons that the entity that possesses them perceives as conferring a benefit if that action is performed.

[pErvinalia]

Scott's definition (which I assume is an analogue of Dennett's).

OK.

Free will is:
The ability of an agent to predict, on occasion, and evaluate the possible outcomes of various alternatives and behave according to that evaluation. This then indicates that there are degrees of freedom; the better an agent's ability to predict and evaluate the more freedom it has. Sometimes, though all the prediction and evaluation in the world will not help if there are no alternatives, in such cases there is no free will. This is the gist of Dennett's argument in Freedom Evolves.

Going back to this definition, that means that my phone has free will because it uses auto-correct.

Free will, as is both commonly understood and is taken to mean in this context, concerns human motives, not merely the existence of alternatives. The clue is in the word 'will', which implies action beyond mere unthinking habit or instinct, actions who's motives have to be distinguished and considered in context.

You are basically shoehorning in metaphysical concepts of "will" and "motive" into a non-metaphysical concept. I judge this to be the case, as you are claiming these are the special sauce that can't exist in software, for example, to cause it to experience free will. Why can't they exist in software? What makes them solely applicable to humans and perhaps animals? Wiki's definition of "motivation" is: "Motivation is a theoretical construct used to explain behavior. It gives the reasons for people's actions, desires, and needs. Motivation can also be defined as one's direction to behavior, or what causes a person to want to repeat a behavior and vice versa." I see absolutely no reason why that definition couldn't apply to software. Can you provide a different definition that supports your position?

The algorithm in your phone does indeed offer you alternatives selected from a base list which is also appended by user interaction - within certain bounds. Predictive text then is wholly deterministic: if you add the same combination of information in you'll get the same information out, depending on a few predetermined variables.

As I said earlier, determinism isn't a rebuttal of compatibilism. In fact, wiki states: "Compatibilist's free will should not be understood as some kind of ability to have actually chosen differently in an identical situation." https://en.wikipedia.org/wiki/Compatibi ... _imaginary

But you, the agent, have to decide which alternative, if any, should make it to your txt, and is goes without saying that your phone does not predict what you intend to say, or when you want to say it, or to whom, or compose and send txt messages without your input.

A hypothetical phone can be modified to do just that.

Your phone has no motives, it has no will. It is not an agent, it is the tool of an agent.

As I said above, this is just injecting some special sauce into humans/animals. What is the metaphysical special sauce?

A phone could be programmed to behave according to one or more motives.

Nonetheless, following your predictive text example, do you consider that which we call free will operates on a similar level or in a similar manner to the predictive txt algorithm, that is; considering a choice of beverages, and given a particular informational state, you'll always and only ever choose coffee over tea, beer over gin, and that this implies that 'free will' is a non-starter?

You are mixing two concepts here - determinism and compatibilism. The two are not directly related. Under the compatibilists definition of free will, I see no reason why software can't be thought to have free will. What the utility of such a definition is, I can't see.

[JimC]

rEv wrote:

...Motivation can also be defined as one's direction to behavior, or what causes a person to want to repeat a behavior and vice versa." I see absolutely no reason why that definition couldn't apply to software...

Agreed in principle, though I doubt it could be usefully applied to existing software, where motivation would not seem to have the reflective basis inherent in human motives, where we can, and do, examine them and confirm whether or not they are rationally in our interests. Neither do current examples of software have the emotional pressures that typically accompany human motivation.

If proponents of AI are correct, that could truly change in future...