Notes

Introduction

1      Cf. on the one hand Cohen (1981) and Dennett (1981), and on the other Bealer (1996a,b), Pust (2001) and BonJour (1998). Though there may be much to admire in these distinct arguments, much in them I would not endorse – see e.g. Lockie (2014b).

Chapter 1

1      Throughout, as should be obvious, I am concerned with ‘epistemic’ rationality (see Foley, passim – e.g. 1993 – for more on this point). Over and above this point, there is the issue of how far and to what extent any philosophical concept of rationality exhausts the open-ended and disjunctive concept of rationality per se; and also to what extent any philosophical concept of rationality may be distinguished from epistemic justification. I shall talk more of these issues anon, but to state my stance very early on: I make no claims that the notion of rationality adumbrated here and in what follows exhausts the concept of rationality per se (and I doubt if any laundering of the concept or other philosophical research project will do this); and I make no claims that it corresponds to any ‘ordinary language’ conception of rationality. I also identify the sense of epistemic rationality I am operating with far more closely with epistemic justification per se than is currently fashionable, and am unapologetic about doing so.

2      ‘What the old mind cannot do – and this is a big limitation of animal cognition – is to make decisions by reasoning about future consequences. That now leaves us with the question of whether epistemic rationality is a useful concept when talking about the old mind, and I rather think it is not’ (Evans 2014: 138).

3      Many accept but many deny this. For one example of the latter, Ryan (2003: 48) distinguishes ‘(1) the view that we have epistemic obligations, (2) epistemic deontologism, and (3) internalism’. Be clear: no one should doubt that some philosophers use (1) the language of epistemic or ethical obligation (‘oughts’, etc.) without drawing on (2) the full deontological framework (OIC, etc.) underpinning these terms. Even more indisputable is (3) the fact that there are numbers of non-deontological (pure access, or mentalist) internalists. I, however, do elide Ryan’s (1), (2) and (3); and I argue in this chapter and (especially) Chapters 2 and 6 that I am right to elide these. There is philosophical disagreement but no confusion here; and the issues are not merely stipulative or (insubstantially) definitional.

4      It is noteworthy that the thin deontological conception of the internalist–externalist distinction was developed as much through the work of externalists (e.g. Alston, Plantinga and Goldman) as through the work of internalists (e.g. Chisholm, Foley and early BonJour). Foley and Plantinga simply embrace this notion of our distinction, or something like it, from the standpoint of (respectively) a proponent and an opponent of such an internalism; and Alston’s arguments were crucial in motivating this distinction (see Chapter 2). Chisholm endorses something very like this notion at a metaphilosophical level, and was the single biggest influence on its modern development; but his specific first-order account (his very complex and involved introspectionist foundationalism) is precisely not such a ‘thin’ concept. Goldman and BonJour (and many others) say things that clearly indicate their endorsement of such an account of the distinction at some high-order level, yet say other things elsewhere (particularly Goldman) that are not compatible with attributing to them a consistent deployment of such a thin, not purely access-based account, especially at a first-order level. Early BonJour (1985: 8, 41–5) accepted this deontic view of internalism, but by 2003 (e.g. pp. 175–7) had recanted – for reasons we consider in the chapter to come.

5      Some philosophers seem to think that internalists (or at least some internalists – Goldman (2009) calls them ‘existential’ internalists, Weinberg (2006) calls them ‘strict’ or ‘absolute’ internalists) maintain that all justifiers are internalist: such that if any justifiers were externalist, internalism, or at least this kind of internalism, would stand refuted (e.g. Greco 2005: 258). I should like to regard this as a straw man, but inconveniently for me, BonJour (2010: 35) appears to embrace this position (though see BonJour 1998: 128–9). However, it is exceptionally and uncharacteristically strong. I am in widely shared company in repudiating it as in any way definitive of our distinction – and certainly as definitive of my position.

6      A strong point worth registering against mentalism as such is that it motivelessly requires us to employ a metaphysical criterion for what is meant to be a key epistemological concept (cf. BonJour 2010).

7      Wu rejects consciousness as central to an account of attention. For Wu, attention is selection for action – including (especially) mental action.

8      Of course, one famous perspectivist is at least substantially an externalist: Ernest Sosa. However, (a) Sosa’s perspectivism is plausibly an internalist component of his account and (b) Sosa’s perspectivism is only terminologically related to the matters addressed here. Richard Foley (e.g. 1993) is a much better example of the type of strongly agent-relative, reasons-based, justificatory perspectivism I am referring to here (see also Alston 1989b). See Lockie (2016a,b) for more on perspectivism and deontology.

9      A minor disanalogy with ethics: the various specific consequentialist theories in ethics differ more over which goal state they specify – utility, preferences or interests – whereas the various specific externalist theories in epistemology agree rather more on the goal state they specify (generally truth, of beliefs, with only occasional diversions into pragmatic utility) and disagree more in the measure of goal maximization they employ to gauge whether this state has been achieved (statistical, counterfactual, causal, etc.). In what follows, I shall use ‘truth’, ‘truth conduciveness’, ‘truth maximization’, etc., as marker terms for the externalist goal state, without regard to the possibility of disputes elsewhere about the specifics of this goal state. (Note that these dummy terms can anyway be allowed to include under their auspices the notion of error minimization, just as the utilitarian’s goal state is tacitly taken to include minimization of disutility as well as maximization of utility.)

10    Cf. Kornblith’s (1983) objections to Boyd and Goldman – that we cannot identify justified belief with reliably regulated belief. With suitable (e.g. time-indexed) provisos, I maintain we can assimilate it to diligently regulated belief, of which more in the next chapter.

11    Putting matters thus enables us to make best sense of early Feldman’s view that the heart of the internalism/externalism controversy concerns whether epistemic value is sui generis (Feldman 1985). Chisholm also adhered to this sui generis view (Chisholm 1988, 1986a: 52). Nottelmann, in contrast, attempts to analyse his basic deontic notion (epistemic blameworthiness), noting of the analysans that the need for it ‘to be non-deontic … should be obvious. If not, no analytical progress would have been made. [Epistemic blameworthiness] would then be flatly circular or analyse epistemic blameworthiness in terms of an equally mysterious deontic notion’ (Nottelmann 2008: 79); later objecting (2013: 2239) that the ‘DCEJ [deontic conception of epistemic justification] must deviate from mainstream epistemology in regarding EJ [epistemic justification] as a non-instrumental good [i.e. good in itself].’ I am solidly on early Feldman and Chisholm’s side here, seeing this as being, at least since Descartes, very much mainstream epistemology.

12    Ryan (2003) rejects OIC – as do Bergmann (2006), Hieronymi (2008), Feldman (2001) and Owens (2000). It is somewhat surprising that Bergmann, the externalist, rejects this principle since, as indicated, it is the classic modus tollens bridging principle to enable the externalist dismissal of a deontic conception of epistemic value – through first rejecting the doxastic freedom requirement of the ‘can’. Ryan (like Steup) is a compatibilist about epistemic freedom, and given how relatively easy it is to defend a conception of freedom, epistemic or otherwise, if one is prepared to embrace compatibilism, it is likewise surprising to see the deontologist Ryan (but not Steup) reject OIC. (Hieronymi’s compatibilism is of a neo-Strawsonian variety and may be supposed to have certain special features thereby.)

13    It should be made clear that this is not intended as a contentful, first-order theory of justification – not in thumbnail sketch as given here, and not as fleshed out and defended in later chapters. For an excellent exemplar of one specific, contentful epistemic theory that is a species of thin deontological account, consider Foley (1993).

14    Deontological approaches are, by some epistemologists, taken to be approaches that are governed by rules. This is incorrect. Deontological approaches are those that embrace sui generis obligations or duties. Being rules-based is neither necessary nor sufficient for deontology.

15    There may be indispensable sine qua non rational principles to some extent, but I suspect these are more for the psychologists than the philosophers to discover (a variable bridgehead, certainly – that’s just a neo-Lucretian argument that you can abandon some rational principles, but not all). Both the Vygotskian tradition in cultural psychology and the Bartlettian tradition in social–cognitive psychology indicate how far from a Meno slave boy are actual, culturally distinct humans from possessing what philosophers would think of as a ‘fixed bridgehead’ of rational principles (e.g. modus ponens: Bartlett 1932; Cole and Scribner 1974; Vygotsky 1978; Lockie 2016a,b).

16    A more technically sophisticated argument than this, defending not only the conclusion, but the OIC premise, is the so-called Anderson–Kanger proof (Anderson 1957; Prior 1957: 140–5). Martin (2009) attempts (I think unsuccessfully) to qualify their conclusion.

17    For an example of a quite brilliant thinker who nevertheless seems to embrace a motiveless access constraint, consider Alston (e.g. 1985).

18    Etymologically, the root term ‘justification’ has deontic connotations from the off, but I am using it neutrally here to mean ‘pertaining to value’, ‘normatively’ or ‘axiologically’.

19    Very much in keeping with this reified approach to deontology, Owens even attempts what might be called a ‘superlative blame’ manoeuvre, claiming: ‘We rightly blame people for things over which they have no control.’ He tries to claw back from the obvious objections thus: ‘but it does not follow that we are entitled to punish people for things over which they have no control’ (Owens 2000: 126). One wants to ask of such ‘blame’ whether it is a wheel that turns yet does no work. If it does some work, call this ‘punishment’ or no, the obvious objections look to resurface (cf. also Owens 2000: 83–4, 141, 149).

Chapter 2

1      Objective and subjective duty because we are here dealing with a response to an objection to ethical/epistemic deontology; shortly, we will generalize this response to a distinction between the subjective and objective without any restriction to duties. I note this because there are clear problems with a (strong) notion of objective duty – at least for one who embraces OIC and seeks after more than an account of the good – cf. remarks about ‘superlative oughts’ in the codicil to the last chapter.

2      Goldman subsequently considers the application of his interpretation of this principle to a specimen ‘rightness criterion’ from the internalist camp – and he chooses as his specimen Richard Foley. He objects that the subjectivism of a ‘Foley rationality’ approach ‘makes Foley’s approach ill suited to the objectivist, nonrelativistic spirit of our entire framework’ (Goldman 2009: 28). But this section has argued such a subjectivism is and must be a feature of any deontically internalist account. What then becomes of Goldman’s claim that ‘this point should be equally acceptable to both internalism and externalism’? Against an internalist of Foley’s stamp, I’d suggest this comes worryingly close to begging the question.

3      Pettit and Brennan (1986) refer to this state of affairs in ethics as one in which the consequentialist conception of the good is ‘calculatingly elusive’ or (more strongly) ‘calculatingly vulnerable’. A common response to this objection (in its many forms) is to look for auxiliary rules – ‘rules of thumb’ – perhaps (Smith 1988) the rule to work from expected utility, or perhaps obedience to the rules of common morality. We will not discuss these responses here; noting, however, that whatever may be said in favour of them on their own terms, there is reason to doubt they will be adequate rebuttals of this point. Two objections that are well discussed in the literature are, firstly, that an expected consequentialism is precisely not a form of consequentialism. A ‘bounded’ restriction of the agent’s justificatory status to expected consequences is motivated by an OIC deontic limit on the grounds for normative appraisal – cf. Clifford’s (1999: 1) judgement on his ship owner: whether a given decision is justified is decided before the consequences are in. Secondly, as regards auxiliary rules/rules of thumb: the objection is made that these lead to regress (Bales 1971; Goldman 1980; Brink 1986; Smith 1988).

4      This is a familiar objection to ‘mentalist’ conceptions of internalism (e.g. Conee and Feldman 2001). Simply put: even exclusively mental justifiers may still be massively inaccessible.

5      I shall use this family of contrast terms strictly as explicated here, and have myself been confused by other authors’ unexplicated usages. Note that this distinction is not to be assimilated to Sosa’s animal versus reflective knowledge – an unrelated distinction that it precedes by many decades. Goldman’s ‘strong versus weak’ justification is clearly somewhere in the vicinity of our distinction (though Goldman (1988: F.N 1) distinguishes that distinction from this). However, I consider the ‘strong versus weak’ distinction to be confusing and ill-formed, accompanied as it is by stipulative and unmotivated entailment claims (e.g. that ‘strong’ entails ‘weak’ – Goldman 1988: 56); claims that represent what I have (in this chapter) entitled ‘halo effects’. There is a double dissociation between the internal and the external success notions in epistemology. Any terminology which obscures the radical nature of that dissociation is to be regretted.

6      The ‘Gordian’, meta-epistemic conclusions of this chapter are conclusions that many epistemologists seem to approach, only at the last moment to veer away. Foley (e.g. 2004) is the one great stand-alone exception here, and I am right glad to acknowledge my debt to him.

7      I have corrected throughout ‘eradicable’ to ‘ineradicable’, as Nottelmann has confirmed (personal communication) that these are typos.

8      Note that I don’t define thin deontologism in terms of blamelessness (cf. remarks on Bergmann in Chapter 1, Section 1.1.3, as well as other remarks in Chapter 5). My deontologically justified subject has typically worked hard for his justificatory status – for instance, to the point where he could be commended for this. But certainly he could be (radically) wrong – conceivably because of ineradicable beliefs. Note also that most justified beliefs (in the epistemic sense at issue) are not ‘punctate’ (e.g. indivisible, sui generis – say, biologically ‘ineradicable’). Usually, even if objectively false, they are a product of articulated reasoning, inferential relations and complex argument. Such argument is (significantly) a cultural–cognitive product, and the beliefs that eventuate from it may indeed be ‘eradicated’ by further argument or diachronic dialectic, albeit not necessarily for an objectively true belief in turn.

9      Newton’s justified false beliefs about physics offer us a clear example whereby, pace Nottelmann, internalist justification is anything but cheaply earned.

10    I mean: the theory advanced by Theaetetus in the Theaetetus (187b), not the theory advanced by Plato in the Theaetetus.

Chapter 3

1      There I contend that indirect, intentionally false belief formation (position 3) is, however, greatly more common. In epistemology (that is, to set aside the distinct literature concerning the alleged paradoxes of self-deception), not very many doxastic involuntarists maintain that direct belief-voluntarism is conceptually incoherent; most maintain that it is (profoundly) empirically false. I maintain that (at least in its stronger forms) it is atypical in non-clinical populations and/or sub-clinically dissociative syndromes and is anyway paradigmatically unjustified.

2      In a tracing case, the person may be culpable or commendable for an action (/cognition) even if he, in virtue of his character, could not have done otherwise at the time of the action (or here, final process of belief formation) provided that earlier free choices of his which ‘trace back’ to this led to the development of his determining character – to himself as he is now. Bergmann and Greco each (separately) see the tracing argument (though neither uses that term) as counting against epistemic internalism (one assumes access internalism): ‘because current blameworthiness can depend on past blameworthiness and past blameworthiness isn’t always accessible, I am not guaranteed to be able to tell on reflection alone whether or not I am currently blameworthy’. (Bergmann 2006: 93; and see Greco 2005: 261). Point 1: I don’t need to be ‘guaranteed to be able to tell on reflection alone whether or not I am currently blameworthy.’ As emphasized in Chapter 1, to be blameworthy doesn’t require a specified (‘tell on reflection alone’) level of access; it requires responsibility-relevant access and control. Point 2: just mark a distinction between synchronic and diachronic (ir)rationality; or if you prefer, synchronic rationality now and synchronic irrationality previously – which point, surely, we would anyway want to mark.

3      The dispute between ethical internalism and externalism is the idiom in which recent generations of philosophers have come to engage with the issue of ethical determinism most generally. Ethical internalists maintain that it is conceptually necessary that moral beliefs motivate (/provide reasons for) moral actions – which claim moral externalists deny. The words ‘internalism’ and ‘externalism’ of course stand for terms taking profoundly different meanings to the terms these words stand for in the epistemic debate.

4      At the time of writing, this paper by Foot was a (rare) example of a genuine occupant of the desires-based ‘hypothetical’ position with which Kant’s maxim is counterpoised. Later, she abandoned that position.

5      Sidgwick (1907) famously criticized Kant on precisely these grounds: that he equivocated between freedom as determination by reason (the Kantian Moral Law – what Sidgwick called ‘Good’ or ‘Rational’ Freedom) versus freedom to obey or break that law (what Sidgwick called ‘Neutral Freedom’ and Campbell (1957) called ‘Elective Freedom’ – to do bad, to be irrational). Note that Kant does so equivocate. Sidgwick then ran precisely this dilemma: either no freedom or no reason. During the same era as Sidgwick, William James soared magnificently: ‘talk of believing by our volition seems, then, from one point of view, simply silly. From another point of view it is worse than silly, it is vile. When one turns to the magnificent edifice of the physical sciences, and sees how it was reared; what thousands of disinterested moral lives of men lie buried in its mere foundations; what patience and postponement, what choking down of preference, what submission to the icy laws of outer fact are wrought into its very stones and mortar (James 1897a: 7). However, James, unlike Sidgwick, by no means rested with this point, nor found these objections compelling.

6      There typically will be considerable differences between agents receiving neuropsychological insult after intact EFs have yielded, say (for Dr. P.), a superior IQ (presumably through acting on other brain structures through the lifespan) versus those who are impaired prior to that (congenitally, say, or through traumatic brain injury in early childhood). (This is true of socioaffective EF impairment also.) EF involves both hot and cold cognition, each of which is significantly mediated by the frontal lobes.

7      Later versions of Shallice’s theory became considerably more complex.

8      Letter-memory task: update WM by remembering the last few words/letters of an ongoing list. N-back task: you are presented with a series of stimuli, one every two or three seconds, and must decide if the present stimulus is the same as one presented n-items back in the list. Flanker task: you must indicate the direction of a central fixation target arrow using a keypad, sometimes when its flanking (distracting) arrows are pointing in the same direction (congruent) and sometimes when they are pointing in a different direction (incongruent – thereby needing inhibition). Go/no-go task: you press a button on each presentation of a letter stimulus as fast as you can – except when the letter is an ‘x’ (when you must inhibit the prepotent response). Trail-making test: connect the circles containing letters and numbers in order, only alternating between connecting by number and by letter (thereby having to ‘shift’). The Wisconsin Card Sort Test (WCST): each card in the deck has one, two, three or four of the same shape on it; these shapes are one of four different kinds (triangles, stars, etc.) and in one of four colours. You have to guess whether you should sort by number, colour or shape and are told only ‘right’ or ‘wrong’. A while after establishing the correct rule, it changes: how quickly can you detect this and change your sorting principle? It is only fair to note that there are longstanding doubts about the psychometric properties (especially reliability) of tests like these; and there are issues as to their short timescale and the fact that they exclusively measure cold cognition. EF is commonly measured by tests like this, but, constitutively, it is a lot more than the cognitive processes subserving only these things.

9      Immediately applied gratification delay (resisting the cream cake) would be an example of impulse control (an inhibitory EF) – or perhaps ‘willpower’ to the extent this is something different. Prosthetically/functionally applied gratification delay (not shopping when hungry, not stocking your refrigerator with cream cakes, avoiding stimulus cues by covering tomorrow’s tea party’s cream cakes from sight in the refrigerator) would be an example of metapsychological regulation; also in this case a (higher-level) inhibitory EF to some extent, but one involving other EFs like WM and theory of mind and umbrella-term EF’s like ‘planning’ – besides, crucially, operating over a much longer timescale.

Chapter 4

1      Quite a few (neuro)psychologists, especially those influenced by factor analytic approaches, do separate out something like three superordinate factors in this fashion. It originated with Friedman et al. (2008) and Miyake and Friedman (2012). I have illustrated a version of their model (similar to that which is endorsed by Diamond) in Figure 4.5. Some distinguish a fourth: specifically, verbal fluency (sometimes ‘self-directed speech’). Best et al. (2009) have as a fourth ‘planning’ (to include ‘problem solving’). Others distinguish ‘dual tasking’ or ‘interference control’ (Willcutt et al. 2005). However, many don’t take such a ‘superordinate grouping’ approach, and the debates about whether such superordinate factors are psychologically real or correlational artefacts are well worked in, for example, the individual differences literature. (Exploratory factor analysis, for all its complexity, is only a descriptive statistic, and the factors extracted are very strongly determined by several arbitrary matters under the control of the researcher – here, notably, the fact that the inputs to this analysis are exclusively short-duration, cold-cognitive tasks.) Note that for any of these superordinate factors there is the problem that they may be too broad and inclusive to be seen as more than expository grouping variables, and too disjunctive at the process level, which is anyway little understood. Also, as already emphasized, finding one putatively orthogonal factor (‘WM’) requiring or logically presupposing the actions of another (‘inhibition’) is a phenomenon that is simply ubiquitous. In contrast to Diamond’s confident assertion, Barkley (2012: 8) strongly repudiates any claim of consensus here, talking of the ‘dog’s breakfast of constructs that EF has become’.

2      Some researchers prefer to see this as simply a tough measurement problem (‘task impurity’: the problem that a given operationalization doesn’t measure only the one EF, or even only EFs rather than lower-brain functions) – a problem to be overcome either with more careful experimental isolation of one’s dependent variable or with greater statistical sophistication (e.g. the use of latent variable analysis; Friedman et al. 2008; Miyake and Freidman 2012). But the low process–behaviour correspondence for EFs (that similar behaviours may have quite different causes) may not be a source of error, (‘task impurity’), being due rather to features that Luria (say) would have immediately understood: the fact that executive functioning is intrinsically holistic, functional, systemic and coordinated. There are lessons to be learnt here from the Bartlettian objections to the Ebbinghausian tradition in memory research: ‘purify’ your DVs too much and you may well throw the baby out with the bath water. ‘Task impurity’ may, then, not be a methodological problem, but a profound functionalist, holistic truth about the in vivo higher cognitive functioning of the human mind.

3      It is not only an elastic notion, but also (like so many of these EFs) has tendencies to be somewhat homuncular – something Baddeley honestly acknowledged of his ‘central executive’ from the start. I am concerned to identify scientifically respectable, roughly agreed truths about the extent of our psychological powers: descriptively accurate accounts of our abilities that any normative epistemologist may make appeal to, and with which any acceptable normative epistemology must comport. I do not then feel that such matters, however much they may be of concern to psychology or even to the philosophy of mind, vitiate my deployment of this and cognate notions, nor require much further engagement in this (epistemological) context of inquiry (although homuncularism I engage with further below).

4      Using TOTE to illustrate WM is borrowed from Conway et al. (2011), but the driving example is my own and is introduced for three reasons, the first being its ecological validity: weak executive functioning is massively predictive of poor, reckless, accident-prone and inconsiderate driving. The second is that it foregrounds variegation in scope as a key feature of both normal and impaired EF: some of these driving problems are more narrowly cognitive (cold cognitive: Norman 1981; Reason 1992 on field dependence, etc.), some more cognitive–affective, some more personality-mediated or agential. The clinicians emphasize things that the cognitive psychologists insufficiently emphasize: that hot cognition and the socioaffective nearly always interpenetrate cold cognition when it comes to executive functioning. Finally, driving is one of the classic examples used to indicate that what counts as executive functioning is not static: what was at one stage an EF becomes, through processes of automaticity, a more posterior cortical function and eventually a lower brain function. This latter is of importance when we reach the section dealing with two-process issues below.

5      This posterior-to-anterior ascent in hierarchical abstraction in rules and reasoning is found in the DMPFC and DLPFC, which have major anterior-to-anterior and posterior-to-posterior connections with each other. There are claims that something like this posterior-to-anterior ascent in abstraction is also found in the ventro-medial PFC (VMPFC; Bechara and Damasio 2005), with the VMPFC occupying a crucial role in mediating social and moral cognition. It is highly likely that these three key areas in the PFC play a major role in deontic reasoning, with there already being evidence that VMPFC-lesioned patients are more utilitarian and less deontic in their judgement of classic (ethical) trolley-car cases (Greene et al. 2001; Greene et al. 2004; Greene 2007; but see Moll and de Oliveira-Souza 2007) – presumably because their judgements are solely informed by consistency and other types of ‘cold cognition’ (as mediated by the DLPFC) rather than also by the ‘hot cognition’ mediated by the VMPFC. If there is a deontic aspect to our learner driver’s invocation of checking procedures, then I would warrant a role for VMPFC. To the extent this remains reasoning, I would expect, however, that it involves VMPFC in close systemic interaction with DLPFC.

6      Jung and Haier’s ‘parieto-frontal integration theory’: sharing resources across a network of regions predominantly in the parietal lobes and frontal lobes in an integrated fashion – cf. Haier (2017).

7      The separation per se isn’t (here) claimed to be illicit: insisting sans motivation that the deontologist must be committed to defending voluntarism with respect to the avowedly automatic ‘final process’ is; while insisting/assuming (sans argument) that the deontologist’s actual, expressed defence of deontologism of the processes leading up to these somehow doesn’t answer to her specifically epistemic needs. Elsewhere (in Lockie 2014b and a planned monograph on knowledge) I ‘problematize’ certain lower-to-higher stage separations of this nature per se (in arguments referring to historic disputes of Bartlett versus Ebbinghaus, and Titchener versus the Gestaltists – the idea that the putatively lower level is unvarnished and prior data: prior, that is, to the higher-level, top-down putative ‘contamination’ of that data).

8      I have too little space to address issues pertaining to personal/practical cognition with the depth they deserve, but awareness of self across time (‘autonoetic awareness’) is a distinctively human, late-developing capacity that involves/integrates with controlled reactivation of memories, planning over variable time-and-distance frames, capacities for ‘decoupled’ (hypothetical) thought and model building involving self-relevant counterfactuals over time. Can such cognition, involving ‘prospection’ (Seligman et al. 2013), be deontically appraised? Yes, in many instances. Does such cognition involve agential-level control? Yes, paradigmatically. ‘The capacity to imagine a hypothetical future from an experienced past is one of the three most important or foundational EFs. The other two are self-directed attention (self-awareness) and self-restraint … together they create the human sense of the future’ (Barkley 2012: 85). Can we deontically assess an agent (as commendable or remiss) for their consideration of future consequences (or lack thereof)? Of course we can, and do. It is even operationalized as a named cognitive–personality variable (Strathman et al. 1994). It is partly mediated, anatomically, by regions that are of importance in deontic reasoning more generally (the VMPFC: Bechara et al. 1994, 2005). A deontic conception of epistemic responsibility is often assimilated to a very ‘synchronic’ notion of responsibility (e.g. Foley, passim); but cognitively, I can deploy the EFs Barkley identifies to travel in time and space, and I may be held responsible for failing to do so.

9      ‘Directing and redirecting executive attention to goal-relevant information may be the primary mechanism by which self-regulatory goals are “shielded” from competing goals or other distractions. … According to this view, goal shielding is the consequence of sustained attention to a goal or task and provides an indirect or “passive” form of inhibitory control. … The ensuing mental state in which individuals “zoom in” on the goals they want to achieve may closely correspond to what has been called an implementation or action-orientation mindset in self-regulation research’ (Hofmann et al. 2012: 175). Notice the promiscuity/reflexivity motif here: executive attention, WM, inhibitory control, self-regulation, passive, active … one explaining the other and then explained by the other in turn. One could seek to clarify a structured path through such accounts, but I think (conceptually, not empirically) that would be unwise. The functional, holistic interpenetration of EFs in vivo is, I think, a profound and important truth about the higher cognitive powers of man.

10    Will this moment of insight then be defined into our involuntarists’ ‘final pathway’ for non-sensory cognition? I take it any such move would simply represent a determination to find a terminological sop for a substantial philosophical defeat. Note that Heisenberg will work forward (in his controlled, executive cognition) from this insightful discovery also (into further mathematics) and will have developed the mathematical expertise that permitted this moment of insight significantly because of his preparedness to work as doggedly yet flexibly as this for many thousands of hours previously over the years; and (more proximally) for many scores of hours previously on this specific problem. His thought is massively diachronic, is pervasively and systemically the product of many different perception–action cycles (Fuster 2013). These cycles, occurring at every level of cognition from the lowest to the highest, involve afferent and efferent processes partly tasked by the agent and partly (as diachronic, developmental and self-organizing processes) leading to the development of the agent, ontogenetically and aretaically. No intellectually serious thinker working on these issues doubts that automatic and uncontrolled cognition is bound up with creative, controlled, executive cognition – or that (in unimpaired agents) each is present in a single mind: cf. remarks to follow on two-process theory.

11    In scare quotes because creativity as an agential-level property is a product of all the EFs and much more besides.

12    We deal with an important update to Maier’s two-string experiment below. Maier’s subjects had been given the problem of tying two strings together that were suspended from a ceiling. The strings were long enough to reach each other if the participant could reach both, but the participant could not stretch to reach the second string whilst still holding the first. A range of objects were in the room that the participants were permitted to use. The correct solution was to tie one string to a pair of pliers then use this as a pendulum to swing this string towards the first until able to grasp both together. This task was difficult for participants because they were functionally fixed on the pliers as a gripping tool – not a weight. This joins several other Gestalt cases as a classic investigation of ‘insight’ problem solving, which in our modern terms mostly concerns flexibility of thought.

13    Miyake and Friedman (2012) see cognitive flexibility (‘shifting’) as involving shifting-specific EF and general EF (= cognitive inhibition, shared by all EFs) – see this chapter’s Figure 4.5. They claim, however, that shifting-specific EF shows some evidence of negative associations with certain developmental indicators of cognitive inhibition (which latter they see as a general factor partly underpinning EFs) – better inhibitory control, to some qualified extent, predicts poorer shifting-specific EF. On reflection, that’s kind of intuitive, isn’t it? Within the individual differences literature, Carson et al. (2003) established that decreased latent inhibition (the ability to ‘gate off’, attentionally, task-irrelevant stimuli) is associated with increased creative achievement (though in high-functioning individuals, not generally – where it may predict psychopathology). Decreased latent inhibition is also known to predict the exploratory cognitive virtues more generally, such as Big 5 O (openness to experience) – in Carson et al., ibid.

14    Whether this token would be an instance of a type – that is, whether EF selection is a sufficiently disjunctive, holistic product of an open system as to be not really (in the sense intended, in a sufficiently meaningful sense) algorithmic (fixed output for fixed inputs thereby mapping a finitely writeable, time transfer-invariant function) is of course a very well-worked debate in classic (functionalist) philosophy of mind. Whatever problems are held to attend functionalism on other grounds and in the context of other philosophical questions, I see no reason to retrench from that position’s standard emergentist stance in this domain.

15    In a sense, we do just this. We are the only organism with language and the only organism with (in any meaningful sense) culture (and technology). We are also, as noted, the only organism with anything like our level of autonoetic consciousness: our sense of self and self-relevant (counterfactual) model building (planning) across time. The socio-cultural extensions of frontal lobe function are extraordinary. Just as the extended mind hypothesis extends mind into the physical world, so the neo-Vygotskian tradition extends mind into culture, education and society: ‘One startling observation Premack made was that chimps do not engage in pedagogy. The flip side of the observation is that humans are the only primates that teach their young’ (Gazzaniga 2013: 12). Clark and Wilson note of Sue Savage-Rumbaugh’s bonobo, Kanzi, raised via a humanly imposed pedagogy using a symbolic keyboard: ‘The system that Kanzi plus his keyboard constitutes forms a cognitive system with memory and other cognitive capacities that seem qualitatively distinctive from that of other, unaugmented bonobos, capacities that are somewhere between those of humans and other apes. It is not simply that Kanzi’s enriched learning environment has restructured his neural wiring (although it has almost certainly done that too), but that his cognitive restructuring has proceeded through a potent cognitive extension involving these stable symbolic structures in his environment’ (Wilson and Clark 2009: 67).

16    ‘Motor structures’ for Fuster, after Koechlin, range from the most basic aspects of muscle movement mediated by the motor cortex, to the highest, most abstract, pure, agential cognition (e.g. Heisenberg cases) mediated significantly by the most anterior areas of the PFC (not intrinsically involving actual motor movement at all). These structures are seen as ranging hierarchically, from those involving the most basic limb movement to those involving the most sophisticated levels of abstract goal-directed thought, but in every case involving multiple self-organizing, open-system feedback cycles from the perceptual to the motor and back.

17    Of course, this doesn’t vitiate the importance of the PFC in EF, or cognitive agency per se. A strong statement of the role of the PFC as ‘capital C’ Controller is the frontal executive hypothesis – that all executive control processes are subserved by the frontal lobes (this is now known to be too strong). Less-strong would be parieto-frontal integration theory (P-FIT) – that attentional and executive processes are shared in a highly integrated fashion between frontal and parietal lobes. Probably matters are more complex than that, too. Nevertheless, even approaches conceding more to the holistic and emergentist traditions in neuropsychology acknowledge the critical role for the frontal lobes in the highest, most specifically human, rational, executive, metapsychological, reflexive, insightful, flexible and controlled thought. ‘My PFC is not my “center of free will” but it is the neural broker of the highest transactions of myself with my environment, internal as well as external’ (Fuster 2013:10).

18    The others are (3) attention regulation, (4) cognitive change/reappraisal and (5) response modification/modulation/suppression.

19    Spacing: the extraordinarily powerful ‘spacing effect’ – distributing practice a little and often and not ‘massing’. Interleaving – introducing ‘contextual interference’ (e.g. practicing different swimming strokes (and different components of swimming strokes) in the same training session and not ‘blocking’). Generating (e.g. generating answers, self-testing, elaborating on the material).

20    This is a slightly under-qualified statement. It may be read as rather too close to Giacomo, Rizzolatti et al.’s committed Premotor Theory, the truth of which is still in the balance (cf. Wu 2014). As regards attention, Wu (2014: 66, emphases in original) usefully distinguishes ‘three possible formulations of the Premotor Theory: the identity, causal and anatomic formulations. The strongest thesis is the identity formulation: the neural circuits for preparing eye movement to location L just are the neural circuits for visual spatial attention to L. The causal formulation is weaker in that it endorses only a causally sufficient condition: normal preparatory processing for eye movement to location L is sufficient to cause visual spatial attention to L. Finally, the anatomic formulation holds that a brain region contains circuitry for preparatory eye movement activity and for spatial attention’. Suitably qualified, however, it is widely accepted that (in some sense) both online and offline efferent circuitry in the frontal eyefields is in intricate connection with (other) frontal lobe functions, in which role they mediate movement of thought.

21    Piaget (1952) contended that, for him, thought was impossible without writing. Wittgenstein absurdly, but rather splendidly, over-stated the Piaget–Feynman thesis thus: ‘this activity [thinking] is performed by the hand, when we think by writing, by the mouth and larynx, when we think by speaking (Wittgenstein 1958: 6).

22    To see how hard the notion of mental action may be pushed, certain brain scientists now take talk of ‘mental agility’ literally, this being a variable mediated by structures (especially the cerebellum) that mediate motor agility. Damage these structures and dysmetria results, but this may be a motor dysmetria (ataxia) or dysmetria of thought – an impairment of mental agility (e.g. Schmahmann 2004).

23    Notably, Johnson-Laird agrees with Evans here. Evans assimilates this (WM) to issues of control and volition. In seeking to understand the Type 1/Type 2 distinction ‘[w]e should not rely on the distinction between conscious and unconscious processing, as both old and new minds have aspects that are conscious and unconscious. … The key difference is that the old mind operates through automated Type 1 systems that have mandatory outputs, whereas Type 2 systems are in some sense volitional: the new mind is capable of forming plans and carrying out intentions under controlled attention’ (Evans 2014: 133, emphasis in the original).

24    Note of their ‘bias’ item in the left-hand column: they are referring to ‘bias’ in the Kahneman–Tversky (heuristics and biases) sense: precisely not our sense – cf. next section. Keren and Schul (2009) in their influential polemic against two-system approaches note (correctly) the heterogeneous nature of such dichotomous lists as that offered by Evans and Stanovich (2013); adding to that heterogeneity a tendency among some theorists to associate System 1 cognition with hot cognition versus System 2 with cold. This is not found in the table I have taken from Evans and Stanovich (2013) and would seem to myself to be quite indefensible on its own terms. We have ‘old mind’ subcortical surges of affect, granted (Evans and Stanovich’s ‘basic emotions’); but we most certainly have ‘new mind’ great systems of ethics, normative epistemology and value theory more generally (Evans and Stanovich’s ‘complex emotions’): we have rigorous socioaffective cognition. Our executive functioning both mediates and is mediated by massive cultural scaffolding and unquestionably involves moral thought.

25    This identification (plus a non-essentialism about the binary distinction) allows us to accommodate an influential criticism from Keren and Schul (2009): that many of the processes subsumed under the two-process distinction are continuous, perhaps dimensional, rather than dichotomous or discrete. We have seen already how executive functioning segues into non-executive functioning – as, say, processes that once were mediated by the most anterior areas of the PFC become progressively more posterior and eventually perhaps no longer mediated by the PFC: via automaticity, proceduralization, etc. I am interested in, but not committed to, two-process/system/type theory; but were I to commit to it, I would not be an essentialist about it, and would want to defend it as a heuristically useful dimensional tendency to be kept in mind through psychological theorizing. Considered thus, some of Keren and Schul’s criticisms are against a conception of the distinction that is excessively strong, abrupt and dichotomous. Note that the distinction between executive and non-executive functioning is clearly useful and well-founded whilst not being ‘ontologizable’ in any simply dichotomous way (e.g. the frontal lobe hypothesis is clearly too strong). Functionally, the distinction between executive and non-executive functioning requires the sort of explication we have seen over this chapter, not a ‘definition’ in any sense: it is a well-founded distinction for all that.

26    ‘Bias’ is also used passim by researchers into ‘hot’ cognition and moral thought (e.g. Anderson et al. 1999).

27    Wu (2014) calls this phenomenon as it manifests in attention the Many–Many Problem.

28    Importantly, in light of (e.g. regress) issues shortly to be discussed, one may also bias one’s biases – cf. Burgess et al.’s (2006) ‘gateway hypothesis’.

29    I suspect that a strong reductionism, narrowly and locally and ontologically conceived, may be needed as a premise to make this argument work: ‘I conceive that nothing taketh beginning from itself but from the action of some other immediate agent without itself’ (Hobbes 1962: 271).

30    There are clear parallels with the classic compatibilist literature here. For three hundred years after Hobbes and prior to Frankfurt, one classic compatibilist after another denied the obvious descriptive truth that there were higher-order cognitions and conations – because, in the grip of a theory and seeing a regress beckoning, they felt that doing so would be insufficiently ‘ultimate’ for their purposes: ‘I acknowledge this liberty, that I can do if I will; but to say, I can will if I will, I take to be an absurd speech’ (Hobbes 1962: 246). This die-in-a-ditch tendency simply melted away when Frankfurt placed other resources at the compatibilists disposal – to the point where classic compatibilism is now (almost?) a museum piece. Yet the armchair psychological truths that Frankfurt adverted to (as to there being higher-order intentional states) should have been quite as obvious, descriptively, three hundred and fifty years ago as they are today.

Chapter 5

1      A point originally made of happiness in the ethical literature (Sidgwick 1907: 405–6).

2      Of course, not if, by stipulation of the thought experiment, the demon is held able to manipulate the envatted agent’s active thought processes – his decision-making, his erstwhile, hitherto seen as executive functioning. Here, there are many similarities between the new evil demon problem and the Frankfurt literature (and the need of the latter to avoid begging the question: Widerker 1995; Ginet 1996; Kane 1996 – cf. Lockie 2014a). We know that one can impair individuals epistemically in suchlike ways without appeal to recherché thought experimentation. But if our demon merely manipulates appearances – say, the sensory inputs to said agent (that is, if the envatted brain, by hypothesis, really has powers of cognitive agency, really is an agent) – then my points stand: he (our brain) can be rational.

3      Reader: if your intuitions are too outraged to see ‘epistemic rationality’ tout court as being this subjective a notion, either substitute a prefixed term of art (‘Foley rationality’?) or substitute ‘epistemic justification’. Little turns on terminology.

4      ‘To abandon the desire for control over beliefs is not to cease to evaluate them. What the epistemologist does is to describe, as best she can, the principles underlying our practices of evaluation’ (Owens 2000: 50). Objections: (1) we never will abandon this desire; (2) to evaluate beliefs presupposes control; (3) what of the epistemologist doing the ‘describing’ of said principles and practices? (Compare the reflexive position of relativists describing the processes of social determination of beliefs, ‘practices of evaluation’ and the like – processes that leave oddly untouched the descriptive and evaluative practices and capacities of said relativists.)

5      I have, in previous work, entitled this ‘ineliminability’ move the Paradox of Abandonment (Lockie 2003c).

6      ‘Fundamental’ not ‘foundational’, note. This style of argument has, as indicated, had many commendable precedents in philosophical history; and in recent work has been represented by Bealer (e.g. 1996a,b), Pust (2001) and BonJour (1998). Although sympathetic to their different attempts to deploy transcendental arguments to defend a profoundly normative, ratiocentric conception of epistemology and metaphilosophy, I would not endorse what they separately take the conclusion of their arguments to be. Specifically, I do not see my own use of this style of argument as committing me to any kind of meta-epistemic foundationalism, nor any commitment to ‘immediate’ (conscious? non-relational?) awareness/intuition. This argument rather opposes (negatively) any attempt to eliminate at some fundamental level a deontic basis for epistemic normativity. I shall not, then, advance a constructive account of said fundamental level, or enthrone this basis for epistemic normativity over all others – of which I then seek a reduction or elimination in turn. I do not have a constructive account of said fundamental level (in particular, not in terms of ‘intuitions’) and would be suspicious of one who did – see my first chapter (remarks on ‘no method’) and Lockie 2014b.

7      Compare this with the debate over the claims of various ‘first-wave’ (irrationalist) cognitive psychologists working in the heuristics and biases literature, which, notwithstanding recent apologetics, at least much of the time (at the time) read as the claim that science establishes that human beings are profoundly irrational. Cohen (1981) and Dennett (1981) advanced transcendental arguments (admittedly rather different to that given here) against the possibility of any fundamental result to this effect, and Stich (1990, 1994) attempted to respond. Though I am very sympathetic to the spirit of Cohen and Dennett’s arguments in this debate, I would not accept much of the details of their respective positions – see below.

8      For an indication of where Goldman ended up, metaphilosophically, forty years later, read Goldman (2007). His (1967) quotation already indicates a sensitivity to the issues I am raising here, and thus that he is by no means the best exemplar of the kind of eliminative, irreflexive stance I am targeting: ‘A main theme of naturalistic epistemology is that the project of epistemology is not a (purely) a priori project. But it doesn’t follow from this that there is no a priori warrant at all’ (Goldman 2007: 19 F.N. 5). Goldman, in 2007 no less than 1967, appears to have turned his ‘naturalistic epistemology’ (externalism) into a very reasonable rejection of a totalizing internalism (Goldman’s 2009, ‘existential internalism’) – not itself a totalizing (‘existential’) externalism in turn. Goldman and I are in agreement so long as that is all that is being defended – see Section 5.2.5, shortly to follow. Note: Kornblith (2007) precisely attacks Goldman (2007) for what he calls his ‘détente’ in this regard: for making his meta-epistemology insufficiently consistent with his first-order views (in my language: for advancing a ‘non-totalizing’ epistemology).

9      Surprisingly, those from the tradition that Cohen bitterly opposed did and do endorse very similar sentiments here – witness Stanovich and West’s (2000) ‘understanding–acceptance principle’ (after Slovic and Tversky). This latter principle is invoked to defend the judgements of irrationality levelled against (e.g. undergraduate) participants who fall foul of Kahneman–Tversky cases. These researchers claim, in vindication of their judgement of irrationality, that after having said cases explained to them at debriefing, ‘more reflective and engaged reasoners [amongst their participants] are likely to affirm the appropriate model [to endorse as correct the reasoning solution that they fell foul of] for a particular situation’ (Stanovich and West 2000: 651). See Lockie (2016a,b) for a comprehensive critique.

10    There are of course no ‘canonical arbiters of rationality’, and Cohen’s narrow reflective equilibrium is no more plausible than the broad reflective equilibrium of the psychologists, decision theorists, logicians and statisticians that he opposes. Note, by the way, that his narrow reflective equilibrium might not be that of a logician or decision theorist, but it is pretty clearly that of at least a moderately educated, at least normally intelligent (and that means highly decontextualized) ‘man on the Clapham Omnibus’ – it is not, say, the reflective equilibrium of an illiterate Uzbek peasant or Kpelle tribesman (cf. the ‘cultural psychology’/psychological anthropology traditions of Bartlett 1932, Cole and Scribner 1974; Vygotsky 1978; Luria 1979 – of which read Lockie 2016a).

11    Leite identifies this distinction in, amongst others, Robert Audi, Michael Williams, Hilary Kornblith and Gilbert Harman (for an example of this move with regard to the pragmatic turn in such arguments, see Chomsky 1995: 26, 27, 53). Leite also correctly identifies this manoeuvre in numbers of (access and mentalist) internalists (he cites BonJour and Chisholm among others). Here, note the ‘Forth Bridge’ argument of Chapter 1, Section 1.2 and the response to Goldman’s tu quoque of Chapter 2, Section 2.3.1: my preferred species of thin deontological internalism will precisely not make space for a (radical, ultimate) deployment of this distinction since it will not permit internalisms of such unbounded complexity that we need another (‘what should we do in the meantime?’) species of epistemic access to discover whether we have attained their putatively internalist (complex foundationalist, coherentist, etc.) success state.

12    ‘Ought’ derives etymologically from the old English past tense of ‘owe’, and in middle English meant ‘under obligation to pay’. Etymological roots for cognate terms are similarly revealing. ‘Deontology’ comes from the Greek deon, meaning ‘one must’, and what one must do is, variously, give what is due (one’s duty), what is binding on one (one’s obligations) and what one is answerable for (one’s responsibilities). Etymology: Oxford English Dictionary (1989) 2nd edn. Oxford: Clarendon Press.

13    Psychology has its footnotes to Plato also, and this is currently a fashionable view amongst those who embrace two-system (two-process) theories. ‘The proposal is that Type 2 processing is best thought of as an internalized self-directed form of public argumentation, and that it is a voluntary activity – something we do rather than something that happens within us’ (Frankish 2011: 20b). Noaves (2011: 254) argues that this was the standard view prior to Kant: ‘The concept of judgement, for example, traditionally used to refer to linguistic claims made by speakers in the public sphere, is transformed by Kant into the mental act of the understanding involved in the apperception of objects. With Kant, logic no longer primarily concerns argumentation; instead, it concerns the inner mental activities of the lonesome thinking subject.’ Chapter 4 already has remarks concerning the Vygotskian notion of the development of thought through the internalization of speech; G. H. Mead of course also approached this via his views as to the social genesis of the self.

Chapter 6

1      A point I owe to Carlos Moya is that Frankfurt may have demurred from this recently (cf. Frankfurt 2003: 344). However, whatever Frankfurt’s current view, endorsing OIC whilst rejecting PAP was a view he held, and is a view that Kant, Wolf and Nelkin have held and do hold. It is noteworthy that the key recent developer of a neo-Frankfurtian approach does not follow him here: Fischer (1999) rejected both PAP and OIC.

2      For example, the ‘prior sign’ manoeuvre (Fischer 1994: 136ff) – and the tortuous expansions and commentaries thereupon.

3      The distinction between simple and complex has a long history as a compatibilist defence against incompatibilist objections. So, Chrysippus’s ‘co-fated’ response to the Lazy Argument (Cicero 1942: xii 30ff) relied on the distinction between ‘simple’ and ‘complex’ facts/occurrents (cf. Bobzien 2001: #5.2.1.1). Fischer’s own distinction between simple and complex acts is considerably under-explicated and appears to be very thinly motivated – perhaps due to his metaphilosophical aim of simply tracking our intuitions (Fischer and Ravizza 1998: 40). But in general (to endorse the classic argument of Reid 1969: Essay 4, Ch. 6), one doubts whether there is any distinction the compatibilist can draw between responsible and non-responsible acts that is not illicitly ad hoc – that will not tacitly track, ex post facto, intuitions that are alone motivated incompatibilistically.

4      In real life, M. Hue most certainly didn’t have to use this device with the heroic (and smitten) Geneviève (Hue and Southby-Taylour 2005: 213–7).

5      I should have preferred to call this ‘Truth World’ and the former ‘Lie World’, but cannot, as Madeleine’s disposition to honesty in Virtue World₂ will not eventuate in her obtaining the truth – due to the presence of the Counterfactual Intervener. As noted and criticized in Lockie (2008), there are interpretations of Aristotle (e.g. Zagzebski 1996: 248) that would deny me even the right to call this latter ‘Virtue World’ as it does not eventuate in success (the good, the truth being told). If this be correct as an interpretation of Aristotle (and it is certainly an accurate description of many Aristotelians), then in using the term ‘Virtue World’, I should simply claim to be operating with another, more intuitively plausible notions of ‘virtue’ and vice’ (say, the Stoic versions thereof – cf. Annas 2003).

6      This line is that of Carlos Moya (2006: Ch. 2, 62ff, 67–70). However, it goes back well before Moya: ‘“I did not shoot him. I did not pull the trigger. My muscles tightened in response to electrical stimulation.” Here we are partly rejecting the did – it was not a deed I did but something that happened to me. We are also partly rejecting the you – it was not I who had a hand in it’ (Lucas 1970: 6, emphases in original).

7      This point needs emphasis and is in no sense directed against a straw man. For instance, in a recent review, we are told ‘a Frankfurt-type counterexample to PAP purports to describe a situation in which: (i) an agent S decides-to-V on his own, but (ii) unknown to S, there is some factor that would have caused him to decide to V, had S not decided-to-V on his own’ (Widerker 2011: 270–1). No such factor can cause Widerker’s agent, S, to decide to V. If it causes him to act, it’s not his decision. It’s not even his act – cf. material cited above from Moya (2006), Lucas (1970) and Lockie (2014a).

8      Shabo (2010a,b) usefully distinguishes causal (‘determines’) from logical (‘entails’) versions of the Direct Argument. The danger with the former is that they may be question-begging against the compatibilist (or at least that they may lead to ‘dialectical stalemate’ with the compatibilist). The danger of the latter is that they may lead to fussy, clever, ‘logical puzzle’ objections to the transfer principle that are not obviously relevant to the core metaphysical concerns of these discussions. Shabo is of the opinion that causal versions, notwithstanding the threat of dialectical stalemate, show perspicuously the motivations underpinning an endorsement of Transfer NR and the price paid by those who flout these principles. I commend Shabo’s discussions to the reader.

9      There were responses somewhat in the neighbourhood of this from numerous pre-Frankfurt twentieth-century compatibilists: to the effect that freedom is not to be contrasted with determinism, but rather with constraint; that we may obey physical laws, yet still be free inasmuch as we are ‘doing what we want’ – much as we may obey the rules of chess (Ryle), yet be free, etc. The twentieth-century rewriters of Hume had no obvious title to these moves, for the chess player’s intentions were quite as determined as the rules of the game; and what the unconstrained agent willed was as determined as what the prisoner was constrained to do. Given that all acts are as determined as each other, and only ‘one path’ is possible, the unproblematic use of a contrast class (a ‘second path’) of something that is unfree in contrast to free becomes very much the point at issue. (See Reid 1969: Essay 4, Ch. 6 concerning the contrast between free and unfree in a totally determined world.)

10    ‘A single counterexample is enough to refute PAP’ (Hunt and Shabo 2013: 604). No it isn’t. The leeway incompatibilist holds that sometimes people must have been able to do otherwise if they are ever to be capable of being responsible for their acts.

11    This ‘neo-Fregean’ example (red) is originally in Lucas (1970: Ch. 3 – he called it his ‘Ionian’ example). He is cautious as to its applicability. I am less so. Recall, we are talking about a predicate without a counterextension anywhere in the domain. And we are talking about a basic predicate, fundamental to moral thought – not a complex predicate, the atomic parts of which avowedly have counterextensions (so, responding with an adaptation of Hume’s ‘golden mountain’ approach, developed as it was to deal with the related problem of non-existence, is not obviously applicable here). In prior research-talk presentation of this material, Derk Pereboom suggested the example of ‘is identical with itself’ as a predicate possessed of sense yet no counterextension (Tim Crane on a separate occasion made a related suggestion). But ‘is identical’ (with other things) has a clear counterextension – and so this remains a ‘golden mountain’ response. This problem (of universal existence) is the mirror image of the problem of non-existence. This latter is often thought to be a pseudo-problem (e.g. Quine 1948; Devitt and Rey 1991), but the ‘golden mountain’-style arguments that seek to establish this do not in any obvious sense work for basic predicates and across an entire domain – a point the later Quine came to recognize when he abandoned that entire neo-Russellian style of approach for his web-of-belief ‘theory’ theory. Without embracing eliminativism, how do we say of this (a basic predicate) that it (what?) has no existence? (cf. Lockie 2003c).

12    Luther’s own reply against Erasmus is overtly cruel and peremptory: ‘The commandments [of God, that, ex hypothesi, the man in question cannot obey] are neither inappropriate nor purposeless, but are given in order that blind, self-confident man may through them come to know his own diseased state of impotence if he attempts to do what is commanded’ (Luther 1969: 191). That is: if he (the damned man) attempts to do what is commanded mind – if he attempts to obey (not disobey) God.

Chapter 7

1      Dana Nelkin (2004a,b) cites Aristotle (e.g. ‘no-one deliberates about what cannot be otherwise’ N.E. 1139 13–14) in her impressive discussion of deliberative incompatibilism. She partly endorses, as a compatibilist response to such positions, the claim that we only are rationally required to refrain from clearly causally inefficacious reasoning (a locked door), not from causally determined reasoning: ‘The motivating idea … is that deliberation is the “difference maker”; that is, your deliberation is what will explain why you do what you do, rather than something else’ (Nelkin 2004a: 230 and cf. her 2011). I cannot see, however, how this survives the Eleatic point: you aren’t really the difference maker – the difference maker was the distribution of energy at the Big Bang. It is noteworthy that in subsequently expanding on what a ‘difference maker’ is, Nelkin (in 2004a) adopts a highly context-relative (less than realist) ‘contrastive explanation’ stance.

2      Susanne Bobzien (2001: 228) allows that this ‘partial fate determinism’ is one possible and plausible interpretation of Chrysippus; and certainly that this was a widely held pre-modern view. Unlike Bobzien, I have no intrinsic interest in the history of this argument; I am interested in why modern philosophers, who surely cannot embrace this ‘partial fate determinism’, nor yet attribute such to their modern opponents, nevertheless take themselves (on the basis of these ancient arguments) to be able to refute a thesis, ‘fatalism’, with such facility. What thesis are they refuting, and what thesis are they defending? Modern conceptions of determinism look likely to be a good deal closer to whatever a modern, non-straw man, ‘philosophical not literary’ conception of (non-logical) ‘fatalism’ might be than modern determinists are wont to acknowledge. And in regard to this ‘non-logical’ qualifier: be clear that we, in contrast to others in the history of this debate, are specifically concerned with the Lazy Argument as a species of causal, not logical determinism. So, O’Keefe (2010) notes that Carneades developed a species of counter to Epicurean versions of the Lazy Argument that was in precise contrast to both the Chrysippan ‘co-fated’ response and the (attributed, and certainly indefensible) Epicurean view that causal determinism inter-entails logical determinism – thus to require both a rejection of the principle of bivalence and an embrace of positive indeterminism (the famous atomic swerve). In contrast to both the Chrysippan and Epicurean positions, Carneades treated the Lazy Argument as a specifically causal argument, to which he offered the solution of self-determinism – the position I defend in Chapter 9 and have partly articulated already in Chapter 4. In summary: the Lazy Argument should not, for a modern author, be assimilated to either logical determinism or partial fate determinism, and arguments that attribute such commitments to the defender of the Lazy Argument in order to overturn it will (when used against modern authors – and also Carneades) amount to straw man arguments.

3      This is a familiar manoeuvre in the dismantling of classical compatibilist analyses of ‘can’ (Berofsky 2002).

4      This argument clearly walks some distance in the footsteps of Popper’s (1982) response to Laplacean determinism. Popper’s response has achieved considerable notoriety because he has been taken to have conflated determinism (metaphysics) with predictivity (epistemology) – attacking the latter yet claiming a refutation of the former. There is, though, no suggestion his attack on the latter was weak when considered only in its own right – and I am here precisely taking issue with Fischer and others’ defence of a predictiveness-based position.

5      So we don’t need an additional, hidden premise (of transcendental idealism, or, in a more modern idiom, some form of verificationism) to make such arguments work. Transcendental arguments are negative arguments. Here, the target of my counterargument attempts to draw a simple realist epistemology–metaphysics distinction to apply to an area whereby the metaphysics is the epistemology. That this state of affairs obtains is not something we have imported in as a concealed, verificationist premise.

6      This appears to be what Fischer means by ‘guidance’ (as opposed to ‘regulative’) control – though (notwithstanding its widespread adoption) his distinction has always appeared worryingly under-explicated to me. In any event, I would contend that ‘control’ in such a case is a misnomer.

7      Stated more sympathetically than I have here, this appears to be in the vicinity both of what Nelkin (2004a) calls the ‘nexus’ view and also her ‘abstraction view’ (she acknowledges that the distinction between these is not easy to draw). The abstraction view has it that nothing outside of the deliberation process makes one choice or other inevitable. However, under the assumption of determinism, something outside of the decision process indeed appears to makes it inevitable – namely, the Big Bang (also: under any set of assumptions, things outside of the deliberation process ought to impinge upon these processes – so, not necessary or sufficient). And the nexus view has it that ‘[r]ational deliberators must believe, in virtue of their nature as rational deliberators, that they have multiple alternatives from which to choose, where their deliberation is the critical nexus (or decisive factor) among those alternatives’ (Nelkin 2004a: 229). I would contend, though, that rational deliberators who are also determinists ought not to believe that their deliberation is the decisive factor among those alternatives – the Big Bang is. Notwithstanding these brief remarks, I commend Nelkin’s (2004a,b, 2011) serious discussions to the reader.

Chapter 8

1      ‘.. is, will be, and has always been, on all occasions false ..’ I should have been quite happy to say ‘necessarily false’ here, but there is a risk this would be misunderstood. It is powerlessness that transfers through the Consequence Argument, and ‘powerless’ is a modal term – but it is not a formal modal operator (it is formally, logically, possible that we go back in time to alter the start-up conditions at the Big Bang or change the fundamental laws of nature). The Consequence Argument seeks to establish that, were determinism true, doing otherwise than we do would be profoundly impossible for us; and this transcendental argument seeks to reason forward from this. See remarks about philosophical arguments versus proofs later in this chapter and the next, and also the comments about causal versus logical determinism in the preceding chapter. Separately, note that ‘ought’ implies ‘can’, together with cannot, entails not ought [‘no oughts’], not ought not.

2      Emphasis Grünbaum. Consider also Hobbes (1962: 256), Nietzsche (1996: #105), Smart (1961), (Pereboom 2011: 417–18) and passim from determinists of various different degrees of hardness.

3      See also Popper’s (1983) response, and Rudder Baker (1988). For remarks on begging the question generally, see van Inwagen (1983: 102ff).

4      As noted in Part 1 of this book, premise 1 is a conditional; and epistemic externalists, who oppose epistemic ‘oughts’, are usually quite as committed to it as internalists (indeed, they require it for their anti-deontological arguments). Opposition to premise 1 tends, then, to be a response specifically motivated by the desire to defend determinism, rather than by any antecedent, specifically epistemic objections to OIC. In light of the issues raised by the Churchland quotation above, make of this motivation for opposing OIC what you may. Earlier versions of the transcendental argument sometimes deploy what we would now identify as an epistemically internalist premise, but with an ‘accessibilist’ rather than my preferred deontological conception of this internalism – for example, Hasker’s (1973: 178) ‘premise A’: ‘For a person to be rationally justified in accepting a conclusion, his awareness of reasons for the conclusion must be a necessary condition of his accepting it.’

5      Even without directly confronting the global asymmetry thesis in the manner given in Chapter 6, note the following: (I) there are transcendental arguments against determinism which are not epistemic in nature that do not need OIC (see, for example, the conative argument, and the related (Kant–Zeno) challenge that determinism can have no practical consequences – cf. Chapter 7, Section 7.2.3). (II) There are even epistemic transcendental arguments against determinism that do not need OIC (I shall consider one in the final chapter). (III) The specific form of the epistemic transcendental argument that I do give can be substantially preserved on some other, nearby principle – for instance, one of the stronger intentional leeway principles already defended in Chapter 6 and Lockie (2014a). So, (III.i) there is a quicker argument available deriving from ‘ought’ implies ‘control’ – though I think it is too quick to compel those not already persuaded of its conclusion. (III.ii) There is a quicker argument available for those who see epistemic responsibility as entailing some version of PAP (or PAB, or cognate principle), and I should be happy to use the arguments of Chapter 6 and Lockie (2014a) to repair to one such position were I unable to meet objections from the asymmetry theorists. For all these reasons, I do not see premise 7 as vulnerable.

6      Surprisingly, Warfield, later in that paper, makes the very strong assertion that ‘in order to get the incompatibilist conclusion … [the defender of the Consequence Argument’s] conditional proofs must not … appeal to premises that are merely contingently true’ (Warfield 2000: 169). As I point out later, a philosophical argument (in contrast, perhaps, to a ‘proof’) may surely be grounded in the profound impossibility of something’s being the case, regardless of whether this profound impossibility is seen as possessing this or that formal modal status. Shabo (2011) makes a similar point, noting moreover that van Inwagen does not derive a modal conclusion in his version of the argument.

Chapter 9

1      I find plausible the intuition that positive indeterminism, even merely as such, might give us something we would value: an unwritten future (whether we are wise to value that or not). But positive indeterminism merely as such doesn’t in any obvious way enhance our responsibility-relevant control (in fact, it seems to threaten such control), and it is hard to see how it could then be a central component of a satisfactory theory of freedom (though on this point I remain interested in views to the contrary – e.g. Balaguer 2010: 125–6). The ‘into, or over, or at’ clause of course merely gestures at what are important differences of detail between the accounts of Kane, Balaguer, Dennett, Ekstrom, etc.

2      See: McCall and Lowe (2005), Ginet (2002a,b), Clarke (1995), Kane (1996, 2002) and Balaguer (2010). O’Connor (2000), despite not being an event-causal or acausal indeterminist, nevertheless offers perhaps the best of these rather variable-quality arguments; however, in the face of strong counters from the determinists – e.g. the ‘replay’ (/‘rollback’) argument – it is still questionable whether positive indeterminism is necessary for free will, and certain that it is not sufficient.

3      A neighbouring, though I think distinct, point – one more used by those who in their different ways embrace the indeterminist horn of the dilemma argument than by those who see this as a false dilemma – is that chance is not an event; chance is an adjective modifying events. Chance doesn’t cause anything, chance describes a property of the causes of things. (I was prompted to make this point by a talk delivered by Audrey L. Anton.) The presence of chance need not then alienate the producer of an action or event from responsibility for that event. Here, by contrast, I claim that negative indeterminism (not positive chance) is an absence of determination by factors external to the agent, not that it is the presence of determination, albeit determination suitably randomized at some (details-specific) point close to the origin of the action.

4      This commits the position to a rejection of the ‘covering law’ model of causation, certainly (at least) where these are interpreted as strict covering laws (the ‘nomologicity of the mental’) – a principle, however, that one should be right glad to reject.

5      This literature developed in two tranches: firstly, out of the work of the classical emergentists – from (in their different ways) Mill (1843), Lewes (1875), Alexander (1920) and Broad (1925) onwards; secondly, from the more recent ‘special science’ philosophy of mind functionalists – from Putnam (1975, 1980) and Fodor (1974) onwards. These approaches have partially merged, philosophically, in the work of O’Connor (1994, 1995, 2000).

6      This isn’t qualified to ‘determination (not causation)’ in order to make a weaker (perhaps even a neo-compatibilist) statement as to the role of the agent – this is intended to be an intransigent statement of libertarianism. This qualification is made because of the suspicion that talk of subscripted, hyphenated causation may be a category error. It is hard to see causation as ‘hyphenated’ (except as a façon de parler, a ‘transferred epithet’ from the sciences or items under which that causal transaction is typed).

7      This challenge is an ellipsis for an argument (call it the Intelligibility of Reduced Predicates Argument) one McLaughlin (1992: 80ff) considers so bad that he assumes only an opponent of Broad would attempt to make a straw man by attributing it to him. (He cites Ernest Nagel (1961: Ch. 11) in this regard – claiming that by Nagel (1963) he had better arguments.) I am genuinely taken aback by McLaughlin’s judgement here (summary judgement, not argumentative response) – which is precisely contrary to my own judgement and that of many others (Fodor, Broad, Putnam, etc.) in the face of this argument.

8      Ginet’s objection may perhaps have force against Clarke’s former view, his ‘integrated’ attempt to combine agent-causation with an indeterministic (probabilistic) event-causal role for reasons. My own position is not to be assimilated to such a view.

9      The connection between moral internalism (ethical determinism) and logical behaviourism is noted in Lockie (1998). Commitment to such a connection is inescapable for one who claims a necessary (a priori) connection between the presence of a reason and a disposition to behaviour. Arguing for the equivalent a posteriori connection simply commits one to an empirical version of behaviourism (or closely similar associationism) – see remarks below regarding Sechenov, Skinner, etc.

10    ‘The first reason of any human action lies outside of it’ (Sechenov 1942: 136 cited in Barbashin 2008); often paraphrased as ‘thought is not the original cause of any action, the stimulus is’. Compare Hobbes, cited above: ‘I conceive that nothing taketh beginning from itself but from the action of some other immediate agent without itself’ (Hobbes 1962: 271, emphasis in original).

11    ‘[T]he question of the autonomy of our mental life does not hinge on and has nothing to do with that, all too popular, all too old question about matter or soul stuff. We could be made of Swiss cheese and it wouldn’t matter’ (Putnam 1980: 134).

Chapter 10

1      I intend this as an ellipsis for a direct (causal) argument. So, Shabo, responding to a putatively responsible yet determined agent envisaged by McKenna (2008), says: ‘Given that the [causal] relationship [between transition mental states] is deterministic, it is hard to see how he can be any more responsible for the efficacy of his deliberative states (the fact that they are efficacious in precisely the way they are) than he is for those states themselves. And, as I have argued, it is hard to see, as well, how he can be morally responsible for his action if he isn’t morally responsible for its causes or the fact that those causes must be efficacious in precisely the way they are’ (Shabo 2010a: 417). See also the quotation by Reid (1969: 266, Essay IV Bk 1) cited below; and note that this direct argument and the various arguments from ultimacy (e.g. from Ultimate Responsibility: Kane 1996) are not mutually exclusive.

2      Strictly, these are terms of art within Kane’s (1996) event-causal libertarianism; I am removing them from the structure of his theory and employing them for my own purposes.

3      I mean this judgement to apply to Nietzsche’s views on free will, eternal recurrence, etc., regardless of whether he is himself esteemed as a mesh theorist; though there are plausible interpretations of his position on free will that, without using this terminology, read very much as if we should take him as such (e.g. Gemes 2006). As regards Nietzsche on eternal recurrence, a representative illustration of the problem that confronts him here would be exemplified by Danto’s (entirely ingenuous) commentary that ‘Nietzsche thought eternal recurrence was a test, and that if we could pass this test there might be some hope for us. If we could, in the face of knowledge he presumed scientific and certain, continue to act despite having acted in just this manner countless times before in periods of the universe just like the present one, which itself is a phase that will return again and again, then one will have achieved a certain meaning in one’s actions and a kind of moral strength. For the action will be seen as done for its own sake (Danto 1987: 203).

4      Also, what is the mark that any such mental content is ‘higher order’? For any given mental item to be higher order is for said item to have a decidedly useful authority for the soft determinist – specifically, the authority to help decide questions as to which of a selection of equally determined actions nevertheless counts as our true choice: historically, a tough, tough question for the compatibilist (Reid 1969: Essay 4, Ch. 6; compare the unsatisfactory response to this challenge in Fischer and Ravizza 1998: 40). For the modern compatibilist – one who has abandoned strongest motive associationism, yet who also rejects incompatibilist self-determinism – specifying what it is to be a higher-order mental item on a non-ad hoc basis (e.g. without tracking ex post facto ‘intuitions’ that are alone incompatibilistically motivated) looks to be by no means merely a problem of analytic detail.

5      Fischer and Ravizza (1998) first coined the term ‘mesh account’. Their own mesh theory is historical (unlike Frankfurt’s) – something that gives them the resources to deal with certain of the manipulation problems (but not all: Pereboom 2001), though thereby has certain costs also. Entitlement to the requirement that historical conditions be satisfied is something that is also open to dispute (van Inwagen 1983: 133). The core problem is that it doesn’t seem to be the presence of an odd history that is producing our revulsion against holding manipulated agents morally responsible (and anyway, what is an ‘odd history’ – in a non-question-begging sense?). It rather seems to be determination by factors not under the agent’s control.

6      In this and any similar position, one sees two other things besides: a metaphysically irrelevant emphasis on a speculated environmental locus of determination – on Bildung – which, in the hands of certain philosophers (and, in particular, many social scientists), is tacitly, indefensibly, treated as if somehow ‘less determinist’ than other loci of determination. And one sees the child’s coerced, ‘educated’ acceptance of his responsibility bounced back at him, but above all at us, his ‘educators’, as a vindicating proof of his ultimate responsibility. For caustic commentary about this latter style of move cf. Nietzsche (1973: Maxim #148) – and of course the famous #21 (‘to absolve God, the world, ancestors … society’, etc.). The hard compatibilist will maintain that the child owns his mechanism without having consented to it (Russell 2010), but the question returns: how, then, may he be held responsible for it?

7      In current philosophy, epiphenomenalism is most often seen as a bogey position, but it has been advanced as a positive solution to the problem of overdetermination; and, as for epiphenomenalism, there are other, classic options here – options that determinists probably need to re-engage with. So, Jordan (1969), in defending his version of the transcendental argument against determinism, suggested (albeit somewhat sardonically) that, in the face of just this problem, the determinist might wish to embrace the notion of pre-established harmony (causal sufficiency for the belief, plus normative epistemic sufficiency in a parallel world of epistemic value). I think actually that determinists might wish to take such suggestions at face value, as attempts to furnish them with a response. In general, given their reflexive situation, determinists should explore the spectrum of such positions, including some of the more rarely entertained, exotic possibilities. For instance, an intellectually serious position that would be available to the determinist here is panpsychism (Skribina 2005) or even pantheism, whereby the ‘World-Mind’ is seen as speaking, rationally, in the determinist – or through the determinist – from 13.75 billion years ago, as to the truth of determinism: a truth that applies to itself. (Though admittedly other avatars of that World-Mind will conflict in this judgement – but perhaps we may allow that it contains multitudes.)

8      His ‘nomological dangler’ objection – which appears to undercut many of the (differently jargonized, but essentially the same) ‘two senses of causation’ positions that followed him (e.g. Yablo’s (1992) distinction between ‘causal relevance’ (of the mind) and ‘causal potency’ (of the quarks); the objection being that if the quarks really are causally potent, the mind is causally irrelevant). Those with a less-than-realist view of causation (Dennett, Rudder Baker) will be unmoved by this objection; but no-one should have a less-than-realist view of causation.

9      Kim’s early work on the exclusion problem (Kim 1989) precisely developed out of his attempt to counter Goldman’s (1969) neo-compatibilist response to Malcolm’s (1968) transcendental argument against ‘mechanism’. Goldman had been taken by many to have developed a strong response to Malcolm; but Goldman’s argument on close re-reading is hard to make sense of and does not appear to survive contemporaneous and later developments in the ontology of the mental. (For example, in his compatibilist response to Malcolm, Goldman (1969: 478) explicitly disavows making a bridge-law/type-type identity connection between reasons and causes – with all the difficulties that at that time were just starting to be seen to attend any such move – yet in invoking something called ‘simultaneous nomic equivalents’, Goldman nonetheless appears to be making exactly such a connection.)

10    Pace Broad (1952: 216) and the odd consensus that has coalesced around the view that his brief, obscure, but highly influential remarks in the essay in question constitute a telling argument against the possibility of non-event causation being temporally located (against, for example, causation by a river, or a self, being located at a time). Clarke (1995) was one of the few who shared my view that Broad’s argument is at best hard to discern, and, if present at all, question-begging; though O’Connor (2011) notes that Clarke appears now to have significantly resiled from this view.

11    As Haji (2008) notes, this way of putting things makes it very close to van Inwagen’s (1980, 1983) original principle (A) in the Direct Argument (or (α) of the Consequence Argument): (‘NRp’ =_df ‘no-one is responsible for the fact that p’.)

12    It is additionally questionable whether the argument above needs to rest on any kind of general transfer principle like van Inwagen’s; and if, tacitly, it does, it is further questionable whether this principle is logical rather than causal (Shabo 2010a,b).

13    This should be seen as consonant with many of the arguments of Chapter 7 – and those of the current chapter’s preceding Section 10.3. It is also of a piece with the objection (e.g. from McKenna 2008) to Frankfurt (‘two-path’) pre-emptive overdetermination counters to transfer considered in Chapter 6, Section 6.6.3: if both paths are assumed to be determined (as they must be), then such cases cannot do the work required of them in non-question-beggingly arguing that there is compatibilist freedom in our putatively one-path world.

14    Cf. Lockie (2003b, 2006). Others, less in vogue now than once they were, would be the Economic Determinism of the Marxists and the ‘Psychic Determinism’ of the Freudians. (Both are remarked upon by several authors in the older literature advancing transcendental arguments against determinism.) I have argued that Eliminativist Materialism has many of the properties of such positions also (Lockie 2003c). If you think that the Platonic and later Greek peritrope arguments work against relativism, yet the Epicurean peritrope arguments do not work not against determinism, ask yourself what the salient differences are here: is it merely the differing locus of determination? What are the important structural differences?

15    I wish to commit to emergentism as the most nearly adequate (/least inadequate) position on special science determination. I remain agnostic on whether to commit to a classic ‘stratified’ model of emergentism.

16    In the philosophy of science literature, this is a point both the reductionists and anti-reductionists substantially agree on – with those who would oppose this being mostly either anti-realists about causation or the defenders of other non-standard views. Remove issues as to the entailments of determinism from the specific area of free will and the compatibilist stance becomes rather more of a minority view. Both Kim and his emergentist opponents substantially agree that any special science causation (real causation – not causal ‘relevance’) would represent a severe problem for the causal completeness of the physical.

17    Campbell distinguishes the ‘inner standpoint’ from ‘the standpoint of external observation’. He assimilates the former to the practical and the latter to the theoretical, though he does (regrettably) assimilate the practical to the experiential – his being an argument from the phenomenological experience of freedom to its reality.

18    I have not in this work articulated my theory of knowledge, and cannot here offer a précis thereof; but I embrace the view that abstract, high-level knowledge is best understood on the basis of modern cognitive psychology’s schema theory – as developed out of the classic approaches articulated by Bartlett (especially) and Piaget and the Gestaltists. On such views, propositional knowledge (at least, high-level propositional knowledge) involves the knower assimilating propositions to very large, intertwined, holistic, inferential, active, top-down knowledge structures – schemata or Gestalten.

19    A commendable exception here is Thomas Nagel, who considers P. F. Strawson’s (1962) quietist views – to the effect that we may properly question only the correct application of the reactive attitudes internal to our practices (say, local determinism due to coercion, mental illness, etc.), and not on grounds external to our practices (overarching metaphysical determinism). Nagel demurs: ‘I believe this position is incorrect because there is no way of preventing the slide from internal to external criticism once we are capable of an external view. It needs nothing more than the ordinary idea of responsibility’ (Nagel 1986: 125, emphases in original). Nagel clearly precedes my own arguments in drawing direct parallels between this libertarian–compatibilist debate and the situation regarding epistemic scepticism, then to note ‘this is a genuine challenge to our freedom and the attitudes that presuppose it, and it cannot be met by the claim that only internal criticisms are legitimate, unless that claim is established on independent grounds’ (Nagel 1986: 126, emphasis added).

20    This should be seen as reinforcing the question of Chapter 7 (e.g. Section 7.2.3 – of Kant and Diogenes Laertius’s apocryphal Zeno) as to whether determinism may be seen as having any practical (here normative-epistemic) consequences at all. Notwithstanding my hostile claims in the introduction, it must be admitted that it does also give us one interpretation of the transcendental argument for freedom in which this is centrally concerned with the ‘possibility of experience’.

21    Anscombe’s response here, in admittedly one of her very earliest papers, may have application against the specifics of Lewis’ (1947) position, but seems less than convincing as a response to any carefully thought-through version of this argument, where it is concerned with the normative epistemology of reasoning (as opposed, that is, to the metaphysical status of its results). James Jordan, though himself a defender of the epistemic transcendental argument, corrected an incautious version of it precisely akin to the Lewis overstatement when he qualified H. J. Paton’s interpretation of the Kantian argument in the Groundwork: ‘Nor would it follow from the truth of any deterministic theory that there is no difference between valid and invalid arguments. What does follow … is that no-one is able rationally to assess an argument for validity’ (Jordan 1969: 53) of Paton (1965) – and cf. remarks in Section 10.4). Consider also the applicability of such remarks to the dismissive hard incompatibilist passage from Galen Strawson (1995: 23) in the previous chapter: ‘one does not want to be self-determining either with respect to … belief or … reasoning, but simply want truth in beliefs and validity in reasoning’.