When this blog launched, I mentioned that part of my motivation was “to ultimately make some claims about the physical world that are relevant to human action”. So far, this territory has been left relatively untouched as I’ve wandered through determinism, emergence, free will, possibility, and the philosophy of mind. But there’s no time like the present. This post will pull ideas from many of those areas in an attempt to partially satisfy that original motivation. In the end, I’ll develop an idea I’ve dubbed “freedom’s debt to reason”.
Preliminaries
I’ll start by assuming the doctrine of determinism, as defined in Determinism and Classical Mechanics, the blog’s first post. Some observations about this assumption are warranted before moving on. First, it’s not water tight, but it’s entirely possible that it’s strictly true. This possibility is left open in certain interpretations of quantum mechanics, namely the so-called “many-worlds” interpretation, and Bohmian mechanics (see Determinism and Quantum Mechanics). Second, it may be that the assumption is, in some sense, true enough. The idea of adequate determinism is that, even though quantum mechanics might describe an inherently uncertain or probabilistic nature at the smallest scales, the phenomenon of quantum decoherence ensures that much of this indeterminacy averages out at larger, but still sub-human, length scales. Thus the behavior of a neuron, at a scale of tens or hundreds of microns, may be adequately deterministic due to the averaging or cancelling out of the quantum effects within it, occurring at nanometer length scales and below. And lastly, even if one is skeptical of the first two points, there are good arguments to be made that the strict truth or falsity of determinism is relatively unimportant when it comes to issues central to the free will debate (on which the line of thinking in this post will partially turn). Strawson touches on these arguments early on in Freedom and Belief. He mentions one line of argument, espoused by others, that
But Strawson advocates a deeper analysis, according to which, 1) “no assumption of the falsity of determinism can help to further the statement of a positive theory of freedom”, and 2) “no objection to a theory of freedom can be raised by assuming the truth of determinism which cannot equally well be raised on other grounds” (7). So whether or not determinism holds is not totally irrelevant. It just turns out to be somewhat unimportant in the free will debate.
That being said, I will still be assuming determinism. I will also be assuming that there is no strong emergence (see Emergence – Philosophical Context). Nothing magical happens to the way systems evolve when they become more complex. In particular, if determinism holds for the underlying low-level physical laws, it will hold at any level of analysis or length scale, regardless of the level of complexity of the system.
Now that we’ve sketched what kind of possibility we want in the ability to do otherwise, let’s outline what types of possibility are consistent with our background assumptions. From an external perspective, viewing the deterministic block universe from the outside, possibilities don’t exist. The block universe just is – all states of affairs at all times are fixed. Abstractly, of course, there are still logical and nomological possibilities (that is, one can still speak of possibility in the block universe), but only one of these possibilities is ever realized – the nomologically possible world with our laws of physics and the initial conditions that actually pertained in the past. The story is very different, however, from an internal perspective. From the perspective of a thinking, knowing being within the block universe, there is a massive amount of epistemic possibility.
For the line of thought being developed in this post, we’ll need to characterize in some detail the nature of this epistemic possibility. It arises from our ignorance, our inability to know, out of a vast set of nomologically possible worlds, which one we actually inhabit. Assuming we know the laws of nature, this ignorance is an ignorance of the precise state of the universe, past or present, and therefore future. An example will be illustrative. For relatively simple systems, our ignorance is small and the epistemic possibility relatively low. We know we’re in the possible world where the next perihelion of Halley’s Comet will occur on July 28, 2061. But there is some uncertainty here – exactly how much, I’m really not sure – maybe seconds or minutes, but probably not hours. Therefore there is a set of possible worlds – one in which the perihelion occurs at 3:38 PM, one where it’s 3:41 PM, and so on – that we cannot accurately place ourselves in. And, assuming we know the laws of nature, this inability to locate ourselves in possible world space stems from our relatively small ignorance about the current state of the system, in this case, the uncertainty in the position and velocity of the comet, among other things. When the perihelion actually occurs, we will be able to locate ourselves more precisely, however marginally, in possible world space by measuring when it happens.
For extremely complex systems, epistemic possibility is very high. Given a chess position, what move will the latest and greatest chess engine recommend? We don’t know. Chess experts could analyze the position using human methods and make educated guesses, probably some pretty good ones. But there’s no way to know for sure what the engine will do without actually running it (in the language of emergence, the chess engine is weakly emergent – see Emergence – Philosophical Context). Only then will we constrain our epistemic possibility, locating ourselves in the set of possible worlds in which that chess engine recommends that move.
To pick up our recurring example, what route will I take to work? We could use behavioral patterns, psychological analysis, and traffic patterns to try to predict the route I will take. Alternatively, you could ask me and assume my response to be truthful and reliable. But like the chess engine, people are weakly emergent – there is no way to know for sure which route I will take without “running” the simulation – observing reality unfold. Doing so will then constrain our epistemic possibility. We now know we’re in the set of possible worlds where I take that route, as opposed to all the others.
Temporal Asymmetry and Freedom’s Debt to Reason
The stage is set – let’s now explore the relationship between the kind of possibility we want in the ability to do otherwise and the kind of possibility that actually exists, given our background assumptions. Essentially, we’re comparing the nomological possibility we want – statement 3), say – with the epistemological possibility we have.
In the present, it seems to me that the epistemological possibility we have gives us at least part of what we want, in a way. When, prior to my morning commute, I say “I could take any route to work today”, the possibilities left open by an epistemic reading of “could” don’t seem to constrain my action. For simplicity, let’s say there are only two routes I would ever take. If these two routes are truly live options (meaning I don’t know anything now that would cause me to take one or the other – I haven’t started the decision-making process), then I do not know which I will take when the time comes to choose. Both routes are open, epistemically. What we want is for them to be open nomologically. My point here is that, phenomenologically, the epistemic and nomological openness might be very similar. True, as I said above, there is an agential quality to the nomological possibility that is conspicuously absent in the epistemic case, and maybe this presents problems for my story. But in both cases, the central phenomenological structure seems to exist – I know of two possible routes, both remain open to me in some sense, I go through a decision-making process to choose one, and I take it. There are obviously a ton of thorny philosophical issues here (see The Phenomenology of Freedom) – I won’t delve into them now. For now I’ll just claim this – in the present, we have the ability to do otherwise, just maybe not in the way we want.
The past is a different story. Here nomological and epistemic possibility are cleaved apart. After I choose which route to take, I observe this, and the epistemic possibility of the present collapses to underlying nomological possibility that was there all along, that is, only one possibility. Observing our own choices forces us to constrain our location in the space of possible worlds to only those that lead to the choices we in fact make. It’s true that past me enjoyed the same overlapping nomological and epistemic possibility that present me is experiencing. But when viewing the past from the present, we no longer have the luxury of our present ignorance. We know which route I end up taking. The epistemic possibility has gone, and with it, our ability to do otherwise.
So we come to see a temporal asymmetry in the ability to do otherwise. In the forward-looking direction, we have it, if only we interpret the possibility invoked epistemically. But this ability seems to vanish when we look into the past because our epistemological situation changes, bringing it in line with the real nomological possibility that was there all along, the lone actuality of the deterministic block universe. It is strange that, according to this account, there is a sense in which it is our present ignorance that grants us freedom. Our observing what actually happens, the choices we actually make, erases that ignorance, and with it, our ability to do otherwise. This is, to invert Ishtiyaque Haji’s concept, freedom’s debt to reason.
It’s nicely reflected in Newcomb’s paradox (see Possibility in Newcomb’s Paradox). Knowing about the cognition simulator, its modelling of our cognition, its predictions of our choices, and how they inform the mad scientist’s manipulation of the $1M seems to constrain our decision-making in a very counterintuitive way. Without this knowledge, we could reason as many do (why take just box B when taking both boxes nets you an extra $1000 no matter what). With this knowledge, we’re forced, on pain of irrationality, to abandon this conventional reasoning, accept that our actions are deterministic, and take only box B. Freedom pays its debt to reason.
Preliminaries
I’ll start by assuming the doctrine of determinism, as defined in Determinism and Classical Mechanics, the blog’s first post. Some observations about this assumption are warranted before moving on. First, it’s not water tight, but it’s entirely possible that it’s strictly true. This possibility is left open in certain interpretations of quantum mechanics, namely the so-called “many-worlds” interpretation, and Bohmian mechanics (see Determinism and Quantum Mechanics). Second, it may be that the assumption is, in some sense, true enough. The idea of adequate determinism is that, even though quantum mechanics might describe an inherently uncertain or probabilistic nature at the smallest scales, the phenomenon of quantum decoherence ensures that much of this indeterminacy averages out at larger, but still sub-human, length scales. Thus the behavior of a neuron, at a scale of tens or hundreds of microns, may be adequately deterministic due to the averaging or cancelling out of the quantum effects within it, occurring at nanometer length scales and below. And lastly, even if one is skeptical of the first two points, there are good arguments to be made that the strict truth or falsity of determinism is relatively unimportant when it comes to issues central to the free will debate (on which the line of thinking in this post will partially turn). Strawson touches on these arguments early on in Freedom and Belief. He mentions one line of argument, espoused by others, that
does indeed constitute a powerful objection to any theory of freedom. Very briefly, it enquires, of any particular event of action, whether it was caused (determined) or ‘random’ (undetermined), and suggests that the agent’s claim to freedom of action is fatally impugned either way. (8)
That being said, I will still be assuming determinism. I will also be assuming that there is no strong emergence (see Emergence – Philosophical Context). Nothing magical happens to the way systems evolve when they become more complex. In particular, if determinism holds for the underlying low-level physical laws, it will hold at any level of analysis or length scale, regardless of the level of complexity of the system.
Now given that determinism holds, and that it holds at all levels, there is a sense in which the past, present, and future are all equally real, all at once. This may sound like an odd thing to claim – let me try to explain. If the evolution of the universe is deterministic, then given some initial conditions, the state of the universe at all future times is fixed. So, given the actual initial conditions of our universe, the states of the universe in the past, present, and future all share a critical feature – they could have been [be, will be] only one way – the way they were [are, will be]. The universe can then be described as simply having specific states at specific times, without reference to past, present, or future. In a sense, they – the past, present, and future – all exist, side-by-side, so to speak. This idea is called eternalism, or the “block universe”. It can also be argued for based on the relativity of simultaneity in general relativity, but I’ve ignored that line of argument for now. It seems to me the block universe follows fairly naturally from determinism alone.
Possibility
It is against these background assumptions – determinism, no strong emergence, and the block universe – that I’d like to dredge up a familiar concept from the free will debate, the ability to do otherwise. My intention is not to rehash the standard disagreements here (see The Freedom to Do Otherwise and Frankfurt Cases for an introduction). Rather, I want to ask something like the following – given a deterministic block universe, what could the ability to do otherwise mean? More specifically, when I say I could have done otherwise or I can do otherwise, what notion of possibility lurks behind the “could” and the “can”? Let’s analyze various notions of possibility (as partially delineated in Possibility) as they relate to the ability to do otherwise and see where that takes us.
First, what kind of possibility do we care about when it comes to the ability to do otherwise? In other words, when we say that we have the ability to do otherwise, what do we want that to mean, and what notion of possibility does it then invoke? Let’s introduce an example that will run through the rest of the discussion – “I could take any route to work today”. When I use the word “could”, what notion of possibility am I invoking? Is it logical possibility? No. I do not merely mean that it is logically possible that I take any route to work today. It’s logically possible that I teleport to Carl Sagan’s house, hop on the dragon in his garage, and ride him to work. But when I say “I could take any route to work today”, I mean to implicitly exclude merely logically possible scenarios that are nonetheless outlandish or fanciful. Logical possibility is too broad and weak a notion to exclude such cases. Is it epistemic possibility? Again, I don’t think so. I don’t mean that I’m definitely going to take route [insert route here], I just don’t know it yet. It’s not that this concept of possibility is necessarily too restrictive. It just doesn’t capture the agency that seems to be implied by “I could take any route to work today” (and which one I take is up to me).
What I want to mean, the kind of possibility I want to invoke, when I say “I could take any route to work today” is something like nomological possibility. It’s worth expounding on this in a bit more detail. By interpreting “could” nomologically, I could mean:
Possibility
It is against these background assumptions – determinism, no strong emergence, and the block universe – that I’d like to dredge up a familiar concept from the free will debate, the ability to do otherwise. My intention is not to rehash the standard disagreements here (see The Freedom to Do Otherwise and Frankfurt Cases for an introduction). Rather, I want to ask something like the following – given a deterministic block universe, what could the ability to do otherwise mean? More specifically, when I say I could have done otherwise or I can do otherwise, what notion of possibility lurks behind the “could” and the “can”? Let’s analyze various notions of possibility (as partially delineated in Possibility) as they relate to the ability to do otherwise and see where that takes us.
First, what kind of possibility do we care about when it comes to the ability to do otherwise? In other words, when we say that we have the ability to do otherwise, what do we want that to mean, and what notion of possibility does it then invoke? Let’s introduce an example that will run through the rest of the discussion – “I could take any route to work today”. When I use the word “could”, what notion of possibility am I invoking? Is it logical possibility? No. I do not merely mean that it is logically possible that I take any route to work today. It’s logically possible that I teleport to Carl Sagan’s house, hop on the dragon in his garage, and ride him to work. But when I say “I could take any route to work today”, I mean to implicitly exclude merely logically possible scenarios that are nonetheless outlandish or fanciful. Logical possibility is too broad and weak a notion to exclude such cases. Is it epistemic possibility? Again, I don’t think so. I don’t mean that I’m definitely going to take route [insert route here], I just don’t know it yet. It’s not that this concept of possibility is necessarily too restrictive. It just doesn’t capture the agency that seems to be implied by “I could take any route to work today” (and which one I take is up to me).
What I want to mean, the kind of possibility I want to invoke, when I say “I could take any route to work today” is something like nomological possibility. It’s worth expounding on this in a bit more detail. By interpreting “could” nomologically, I could mean:
- that there is a set of possible worlds with all the same laws of physics and initial conditions as ours, in each of which I take one of various routes to work,
- that there is a set of possible worlds with all the same laws of physics as ours, but slightly different initial conditions, in each of which I take one of various routes to work, or
- 1) or 2), and my choice of which route to take determines which of these possible worlds is actualized.
Now that we’ve sketched what kind of possibility we want in the ability to do otherwise, let’s outline what types of possibility are consistent with our background assumptions. From an external perspective, viewing the deterministic block universe from the outside, possibilities don’t exist. The block universe just is – all states of affairs at all times are fixed. Abstractly, of course, there are still logical and nomological possibilities (that is, one can still speak of possibility in the block universe), but only one of these possibilities is ever realized – the nomologically possible world with our laws of physics and the initial conditions that actually pertained in the past. The story is very different, however, from an internal perspective. From the perspective of a thinking, knowing being within the block universe, there is a massive amount of epistemic possibility.
For the line of thought being developed in this post, we’ll need to characterize in some detail the nature of this epistemic possibility. It arises from our ignorance, our inability to know, out of a vast set of nomologically possible worlds, which one we actually inhabit. Assuming we know the laws of nature, this ignorance is an ignorance of the precise state of the universe, past or present, and therefore future. An example will be illustrative. For relatively simple systems, our ignorance is small and the epistemic possibility relatively low. We know we’re in the possible world where the next perihelion of Halley’s Comet will occur on July 28, 2061. But there is some uncertainty here – exactly how much, I’m really not sure – maybe seconds or minutes, but probably not hours. Therefore there is a set of possible worlds – one in which the perihelion occurs at 3:38 PM, one where it’s 3:41 PM, and so on – that we cannot accurately place ourselves in. And, assuming we know the laws of nature, this inability to locate ourselves in possible world space stems from our relatively small ignorance about the current state of the system, in this case, the uncertainty in the position and velocity of the comet, among other things. When the perihelion actually occurs, we will be able to locate ourselves more precisely, however marginally, in possible world space by measuring when it happens.
For extremely complex systems, epistemic possibility is very high. Given a chess position, what move will the latest and greatest chess engine recommend? We don’t know. Chess experts could analyze the position using human methods and make educated guesses, probably some pretty good ones. But there’s no way to know for sure what the engine will do without actually running it (in the language of emergence, the chess engine is weakly emergent – see Emergence – Philosophical Context). Only then will we constrain our epistemic possibility, locating ourselves in the set of possible worlds in which that chess engine recommends that move.
To pick up our recurring example, what route will I take to work? We could use behavioral patterns, psychological analysis, and traffic patterns to try to predict the route I will take. Alternatively, you could ask me and assume my response to be truthful and reliable. But like the chess engine, people are weakly emergent – there is no way to know for sure which route I will take without “running” the simulation – observing reality unfold. Doing so will then constrain our epistemic possibility. We now know we’re in the set of possible worlds where I take that route, as opposed to all the others.
Temporal Asymmetry and Freedom’s Debt to Reason
The stage is set – let’s now explore the relationship between the kind of possibility we want in the ability to do otherwise and the kind of possibility that actually exists, given our background assumptions. Essentially, we’re comparing the nomological possibility we want – statement 3), say – with the epistemological possibility we have.
In the present, it seems to me that the epistemological possibility we have gives us at least part of what we want, in a way. When, prior to my morning commute, I say “I could take any route to work today”, the possibilities left open by an epistemic reading of “could” don’t seem to constrain my action. For simplicity, let’s say there are only two routes I would ever take. If these two routes are truly live options (meaning I don’t know anything now that would cause me to take one or the other – I haven’t started the decision-making process), then I do not know which I will take when the time comes to choose. Both routes are open, epistemically. What we want is for them to be open nomologically. My point here is that, phenomenologically, the epistemic and nomological openness might be very similar. True, as I said above, there is an agential quality to the nomological possibility that is conspicuously absent in the epistemic case, and maybe this presents problems for my story. But in both cases, the central phenomenological structure seems to exist – I know of two possible routes, both remain open to me in some sense, I go through a decision-making process to choose one, and I take it. There are obviously a ton of thorny philosophical issues here (see The Phenomenology of Freedom) – I won’t delve into them now. For now I’ll just claim this – in the present, we have the ability to do otherwise, just maybe not in the way we want.
The past is a different story. Here nomological and epistemic possibility are cleaved apart. After I choose which route to take, I observe this, and the epistemic possibility of the present collapses to underlying nomological possibility that was there all along, that is, only one possibility. Observing our own choices forces us to constrain our location in the space of possible worlds to only those that lead to the choices we in fact make. It’s true that past me enjoyed the same overlapping nomological and epistemic possibility that present me is experiencing. But when viewing the past from the present, we no longer have the luxury of our present ignorance. We know which route I end up taking. The epistemic possibility has gone, and with it, our ability to do otherwise.
So we come to see a temporal asymmetry in the ability to do otherwise. In the forward-looking direction, we have it, if only we interpret the possibility invoked epistemically. But this ability seems to vanish when we look into the past because our epistemological situation changes, bringing it in line with the real nomological possibility that was there all along, the lone actuality of the deterministic block universe. It is strange that, according to this account, there is a sense in which it is our present ignorance that grants us freedom. Our observing what actually happens, the choices we actually make, erases that ignorance, and with it, our ability to do otherwise. This is, to invert Ishtiyaque Haji’s concept, freedom’s debt to reason.
It’s nicely reflected in Newcomb’s paradox (see Possibility in Newcomb’s Paradox). Knowing about the cognition simulator, its modelling of our cognition, its predictions of our choices, and how they inform the mad scientist’s manipulation of the $1M seems to constrain our decision-making in a very counterintuitive way. Without this knowledge, we could reason as many do (why take just box B when taking both boxes nets you an extra $1000 no matter what). With this knowledge, we’re forced, on pain of irrationality, to abandon this conventional reasoning, accept that our actions are deterministic, and take only box B. Freedom pays its debt to reason.
