At several points in The Rational Imagination, Byrne (Reference Byrne2005) presents her argument as a three-step (or three-premise) argument to the conclusion that counterfactual imagining is rational. Insofar as this argument is valid, the conclusion is weaker than it seems. Moreover, it does not represent the central contributions of this book, which, if anything, point instead to what is irrational about counterfactual imagining.

1. Human reasoning is guided by rational principles

Byrne begins with the premise that human reasoning is rational – a premise that she explicates as the claim that human reasoning is based on rational capacities or principles in much the same way that human language is based on grammatical capacities or principles. Even though our actual reasoning performance (like our actual speaking performance) is often flawed, that is due to such factors as the constraints on our working memory, not to any flaws in the cognitive mechanisms that define our underlying competence. Here Byrne is aligned with Jonathan Cohen (Reference Cohen1986), among others, who insists that human irrationality occurs at the level of performance, not competence; but she does not offer any arguments for this position.

The main alternative to this approach claims that human reasoning is based, instead, on various heuristics – heuristics that are not rational in the logical, truth-preserving sense, yet are usually useful in the contexts humans find themselves in – contexts in which time and energy are limited, contexts in which some possibilities are much more likely or much more significant than others, and so forth. This alternative has little need for a competence/performance distinction since the specified heuristics are supposed to be nature's way of factoring our limitations into our design. (The analog, in the case of language, locates the grammar of a language in the way people actually talk, rejecting the notion that there is a deeper and more competent grammar that underlies our speech, against which our flawed performance may be measured.) Whether such heuristics constitute a different, or better, type of rationality than that defined by logical principles will determine whether or not advocates of this alternative would agree with Byrne's first premise as stated.

2. Human reasoning is guided by imagined possibilities

In at least some summaries of her argument (pp. 198–99; 215), Byrne's second premise states that our reasoning depends on thinking about, or imagining, possibilities. (Thinking about possibilities is effectively equated with imagining possibilities – particularly when the possibilities in question are mere possibilities. Nothing turns on whether the imagining in question involves imagistic representations.) This is a claim that Bryne and Johnson-Laird have been defending for some time (and one which, as she notes, has a long philosophical lineage). Here the main alternative is a version of the inference rule theory – the theory that our inferences are guided by more or less complex sets of rules that tell us how to proceed from premises to conclusions; that our reasoning consists in applying such rules rather than actively considering a range of possibilities. There are several points in Byrne's book where she argues against this alternative (e.g., pp. 51–52 and pp. 115–16), but these passages are fairly brief and inconclusive. This book does not (primarily) address that dispute.

Note how the above two premises combine in the case of deductive reasoning, according to Byrne:

From what I have described so far, one might expect Byrne's project to be a detailed account of just how limits on our imagination constrain our reasoning performance – how our imaginations select from the totality of logical possibilities in order to produce our (often flawed) reasoning performance. And this is precisely what I think she actually does. Chapter by chapter, she describes particular ways in which we limit the logical possibilities that we imagine when reasoning about what actions or conditions would or would not have made a difference, what ought to have happened, what we regret, and what we deem inevitable. These are interesting, useful, and nicely documented observations, deserving of close attention and continued discussion.

Byrne's stated position, however, is something different, for the conclusion of each version of her three-step argument (and what she repeatedly calls the central idea of the book) is this:

Given the severe restrictions on the sets of possibilities that we are said to imagine when contemplating what would have made a difference, or what we should have done, and so on, and given Byrne's equating of deductive competence with an ability to imagine all relevant possibilities (and thus all possible counterexamples), this is a surprising conclusion.

3. Counterfactual imagining is guided by the same principles as those that guide human reason and imagining possibilities

Byrne's crucial third premise, which is stated in a number of different ways (compare versions on pp. 38, 199, 208, and 215), claims that the principles that guide human counterfactual imagining – which is a subset of imagining possibilities more generally – are the same principles as those that guide human reasoning. Since premise 1 affirms the rationality of these principles, it is now an easy step to the conclusion that counterfactual imagining is rational.

As we have already noted, though, the most that this would establish would be our competence for rational counterfactual imagining; it would not ensure the rationality of our actual counterfactual imagining. And given Byrne's careful detailing of the many possibilities that most of us do not (usually) consider in counterfactual imagining (possibilities that are less controllable, possibilities that are forbidden, possibilities that are in the more distant past, etc.), it is clear that our actual performance falls far short of our underlying competence. If the third premise is understood as referring to the principles that characterize our rational competence, it would seem more appropriate to conclude that counterfactual imagining is irrational – not in principle, but in fact.

On the other hand, if the principles that guide counterfactual imagining are captured in the list of principles that Byrne articulates, chapter by chapter (see Table 7.2, p. 161, for the complete list), then there is little reason to think that these principles (e.g., “People keep in mind few possibilities”) are the principles that are constitutive of our rational competence. They may be efficient or instructive or reassuring in some of the ways that Byrne sketches (on pp. 209–12), but they do not take account of all relevant alternatives. If, when considering how things might have been different, we restrict ourselves to imagining changes in controllable factors only, or changes in only the most recent events, surely we are not fully exercising our capacity for rationality.

If, as seems plausible, human reasoning rightly relies on a combination of logical principles and pragmatic principles, then it is not surprising that counterfactual reasoning and counterfactual imagining also rely on a combination of logical and pragmatic principles. That is not a surprising conclusion, and it is not where the real interest of Byrne's book lies. The most significant contribution of this book is her description of the ways in which certain possibilities are usually not imagined when we reason with counterfactuals – ways in which our rationality is, understandably, limited.