What if John were a fish? I suppose he would be able to breathe underwater. Would he be able to survive the vacuum of space? I think not. The counterfactual ‘If John were a fish, he would be able to breathe underwater’ seems to be true, whereas the counterfactual ‘If John were a fish, he would be able to survive the vacuum of space’ seems to be false. These are more than just counterfactuals – they may be construed as counterpossibles, or counterfactuals with impossible antecedents. They present strange yet familiar pieces of reasoning that form the most distant reaches of our intuitions relating to the nature of our world, setting its most fundamental limits, whether in terms of the very question of the nature of impossibility or the contrast between the apparent truth values of the above two counterpossibles. It is these distant reaches that Maciej Sendłak explores in his Wondering about the Impossible: On the Semantics of Counterpossibles.

The book is an accessible introduction to counterfactuals, with multiple chapters devoted to building out various major, established positions. At the same time, it presents a surprising number of novel arguments, together with extensions of various positions, including a new hybrid account of counterfactuals. Three chapters outline different approaches, namely inferentialism, possible worlds semantics, and truthmaker semantics. The last chapter brings together some of the key ideas developed throughout the book to outline Sendłak’s own hybrid position involving dependence relations and Meinongianism (sic!).

Turning to the initial set of chapters, the book opens with a discussion of inferentialism, ranging from early forms of inferentialism discernible in Mill (Reference Mill1868) and Ramsey (Reference Ramsey and Ramsey1925), through the classic positions defended by Goodman (Reference Goodman1947) and Chisholm (Reference Chisholm1946, Reference Chisholm1955), and finally the most recent accounts due to Douven et al. (Reference Douven, Elqayam, Krzyżanowska, Kaufmann, Over and Sharma2023). Each position is determined to be circular when extended to counterfactuals, as the background conditions and laws that are thought to make the inferences work are shown to depend upon counterfactuals. Sendłak retreads classic arguments along with interesting extensions and modifications, introducing counterpossibles that further accentuate where they go wrong. Of particular note is how the most recent accounts indicated above are approached, in that they are not designed to work with counterfactuals but rather explicitly focus on indicative conditionals. Sendłak develops an extension of those accounts, calling it ‘neo-inferentialism,’ which is in itself novel and interesting. While the extension appears to be well-reasoned and charitable, it leads to similar difficulties involving circularity.

This discussion of inferentialism is followed by Sendłak’s treatment of possible worlds semantics (PWS), which is by far the longest chapter of the book, and also perhaps the most robust in terms of its exposition. He is quick to note that his discussion is not exhaustive, which is not surprising given the breadth of the literature on PWS. That being said, what he does cover is well-chosen and presented. He frames the whole discussion in terms of ‘orthodoxy’ and ‘unorthodoxy.’ The former corresponds to classic positions due to Lewis (Reference Lewis1973, Reference Lewis1986), Plantinga (Reference Plantinga1976, Reference Plantinga1974), and Stalnaker (Reference Stalnaker1976, Reference Stalnaker and Rescher1968); the latter corresponds to more recent developments within the PWS framework associated with the hyperintensional revolution (Nolan, Reference Nolan2014). Putting it briefly, orthodoxy does not make room for impossible worlds and renders all counterpossibles vacuously true, including that if John were a fish he could survive the vacuum of space. This contrasts with unorthodoxy, which accommodates impossible worlds, and so accommodates counterpossibles with different truth values. In developing unorthodoxy, Sendłak makes use of Priest’s (Reference Priest1997) ‘Australasian’ and ‘North American’ distinction of the positions to frame his own discussion. The key to each is that the Australasian approach limits us to worlds in which some logic holds, whereas the North American one introduces no such limitation.

Sendłak outlines the familiar case against the Australasian approach, appealing to examples like ‘If there were no true logic, there would be no true logic’ (p. 74), which are taken as false insofar as the Australasian approach requires worlds to abide by a logic. However, the above does seem to be true, motivating support for the more liberal North American approach, which is associated with the idea that for any given formula, there is a world where it holds and where it does not hold (see, e.g., Laan, Reference Laan1997; Priest, Reference Priest2005, Reference Priest2016). However, Sendłak complicates things by introducing what I consider to be a compelling counterexample; he confronts the North American approach with (PI) ‘This world is both logically possible and impossible’ (p. 75). The issue is that the very framework of PWS separates possible and impossible words, and so it appears to be a necessarily false formula. As such, it should be true at some impossible world; however, it cannot be true at some impossible world, because that world cannot be both impossible and possible. The argument leads Sendłak into one of the highlights of the book, as he introduces a novel approach to unorthodoxy that fits between the Australasian and North American ones, which he refers to as the Pacific approach. The idea is to bring together the notion of truth about a world (as distinct from truth at a world) and metalinguistic considerations, such that the North American notion of any given formula holding in a given world is commuted to any given object language formula holding in a given world. This of course handles the initial problem with PI, while also inviting some further challenges, modifications, and concessions; the way it is developed is interesting and not without potential issues, but the solution itself is certainly impressive in its simplicity, and it will likely be subject to further discussion in the literature. His pacific account is followed by further discussion of different unorthodox approaches, problem cases, with the chapter concluding with an extended discussion of the interplay between orthodoxy and unorthodoxy more generally.

Sendłak then turns to truthmaker semantics (TMS). The first half of the chapter concerns Fine’s (Reference Fine2012, Reference Fine2017) position, whereas the second half presents an original Meinongian account of truthmaking. The discussion of Fine’s position is the most difficult part of the text to follow in terms of exposition, in part due to the technical complexity of Fine’s work, but also in part due to the pacing with which it is introduced. Sendłak establishes that TMS is committed to orthodoxy and moves to consider how it handles Sobel sequences. In doing so, he raises the question of how TMS handles outcomes of fusions, in particular whether fusions inherit the outcomes of their elements. Here we see an argument where the author seems to get it very wrong. Cutting through the technical perspective at issue, we are presented with a case where Amélie decides to go to a few cities and to visit one art museum in each city. The sequence we are given is that (A6) ‘If Amélie went to NYC, she would visit one museum’, and (A7) ‘If Amélie went to London, she would visit one museum’, from which we are to infer (A8) ‘If Amélie went to London and NYC, she would visit one museum’ (p. 137), which Sendłak presents as a bad inference. In contrast, we ought to want to infer (A8*) ‘If Amélie went to London and NYC, she would visit two museums’, which is complicated by the fact that the consequent is not technically among the inherited outcomes. However, it is not clear that (A8) is a bad inference, in that Amélie would indeed visit one museum (in fact, she would even visit two). Further, the problem posited in both (A8) and (A8*) depends upon the choice of language describing the situation, such that we could equally well phrase the consequent as ‘visit a museum in each city’, which is how the case is presented in the first place. This throws some doubt onto Sendłak’s problematization of Fine’s position, though he also raises other concerns independent of this particular case.

In contrast to the treatment of TMS, which leaves something to be desired, the presentation of what Sendłak terms ‘Meinongian abstractionism’ is excellent. After laying out the relevant features of Meinongianism, including the notion of ‘objectives’ that play the role of truthmakers, he marries it with the possible worlds framework, with a set of possible and impossible worlds, accessibility relation, domain, and so on. The details of the account are complex but accessible and surprisingly natural, with the resulting picture being one in which ‘the universe of sentences is not divided into just two groups of true (and true only) and false (and false only)’ (p. 147). That is, the position provides a rich distribution of truth values out of the box, such that a given sentence may be true and not false, not true and false, both true and false, or neither true nor false at a given world. And it is precisely in the context of a discussion of counterpossibles and impossible worlds that a position like this is at its best; where other positions are based upon a narrower distribution and operate out of classical logic, the Meinongian position naturally maintains the sort of scope that other positions need to be extended to. It allows for the intended unorthodox treatment of counterpossibles, and it also works together with the stipulation concerning object-language formulas drawn from the Pacific approach, presenting something very unique and interesting in its own right.

Elements of all of the above discussions are brought together in formulating the hybrid ‘dependence account of counterfactuals’ (DAC), which brings together the motivation found in the treatment of inferentialism, the framework of Pacific possible worlds, and the metaphysics of Meinongian truthmakers. It opens with a discussion of explanation and places dependence as the relevant connection between the antecedent and consequent of a given counterfactual, such that a counterfactual ‘is true at w iff there is a world where the ontological correlates [(objectives) of the antecedent and consequent] are related via some dependence relation, and the same dependence relation holds at the world of evaluation, w’ (p. 176). The inclusion of the dependence relation and its status in the world of evaluation introduces a further layer of complexity that allows for difficult assessments involving various counterpossibles together with varying truth values. It is an account that is in many ways developed around the impossible in equal measure as the possible, and one that appears to succeed in what it sets out to do, presenting a robust and well-articulated treatment of counterpossibles. The complexity of the account, with its many moving parts and influences drawn from different perspectives, is a strength of the book as a whole and speaks to its value, in that it achieves both a broad overview of the counterfactual literature with a unique emphasis on counterpossibles while also advancing a series of original arguments and examples together with the hybrid account that it concludes with.