PLEASE DO NOT CITE. PRELIMINARY VERSION Zuzanna Rucinska Chapter III. The Nature of the Investigation – seminar version 1. Introduction The purpose of this chapter is to clarify the metaphysical commitments behind taking different methodological routes to explaining pretence. The nature of the investigation changes depending on whether one appropriates, for example, a functionalist methodology or an enactivist methodology. In result, the types of explanations one can provide about pretence will strategically differ. What this chapter will involve is the following. Firstly, in subsection 2, I point to a common practice in the present literature on pretence, which is that mental representations get to enter both as essential make-up of pretence, and as explanatory posits that partake in the mechanism of pretence. In what follows I look at the relationship between essences and mechanisms. I challenge the validity of the claim that mental representations must explain pretence even if it is to be described as representational. I conclude that the essential make-up of any aspect of cognition should not influence how it is to be explained. In subsection 3, I engage in a discussion about different approaches to explaining mental phenomena, providing brief sketches of two very different approaches, functionalism and enactivism. Functionalism is discussed because it is one of the most popular, but at the same least demanding (unlike, for example, identity theory), approaches, which uses mental representations to both explain and constitute mental phenomena like pretence. It will be compared with enactivism, which is the non-representational approach I follow throughout the dissertation.1 This subsection focuses on how mental representations get introduced in the explanatory 1 While it could be, in principle, possible to argue for functionalism without mental representations (though it has not been practiced), and although some who claim enactivism do make use of a weak notion of mental representations (for example, see Scarinzi’s critique of O’Regan’s enactive sensorimotor approach in Bishop and Martin (2014, p. 71), hereby for simplicity sake I will equate functionalism with a mental representational account and enactivism with non-mental representational one. 1 story of a functionalist: as forming the essence of pretence, and as forming their implementation mechanisms. On the functionalist story, external factors are treated as causes and are separate to the role-realiser structuring of pretence. A comparison to enactivism will be made, which has a different approach to explanations: by following the dynamical systems theory, the distinction between ‘synchronic’ implementation mechanism and ‘diachronic’ causal factors get blurred in enactivism. Then, in subsection four, I hone in the question of providing mechanistic explanations. I discuss the mechanistic nature of the cognitivist (including functionalist) explanations, and show that, while commonly used to explain various phenomena from perception (Marr) to empathy (Gallese), they may not always work. I will also provide examples from enactivism to explaining perception (Varela et al.) that do not lean on just taking a mechanistic approach to explaining colour experience. What follows is a discussion on the types of strategy enactivists can take: either endorsing a ‘no-mechanism’ view that says that proposing an explanatory mechanism of pretence cannot be appropriated by enactivists, or a ‘wide-mechanism’ view that tries to incorporate social and environmental aspects into the explanatory mechanism (without changing the explanandum in the relevant sense). I propose that, ultimately, non-representational explanations of enactivists should not follow a mechanistic structure. With subsection five I conclude with possible objections of a cognitivist interlocutor to the question of whether the enactivist is proposing a genuine alternative explanation, or whether or not the enactivist changed the target of investigation. I provide enactivist rebuttals to both worries, engaging in a brief investigation into the nature of philosophical explanations and the question of what is the role of an enactivist philosopher. I will clarify that functionalism and enactivism have different ways of explaining what seems to be the same target – the phenomenon of pretence. However, as proponents of representational theories take pretence to be constitutively a mental representational phenomenon (with features such as imagination or knowledge), they target enactivism as a theory that is targeted at a different explanandum. In that sense, both theories are possibly compatible. However, I will argue that the explanandum is, or at least could be, the same (broadly construed pretence phenomenon that has the same features like knowing and imagining, but which are described non- 2 representationally), and thus, the theories cannot be compatible, and enactivism is a relevant counter approach to functionalism. This chapter is an important addendum to the thesis as it clarifies the overall dialectic of the thesis; it brackets some worries with seeking non-mental representational explanations of pretence, and clarifies the reason why certain working assumptions will be taken to provide such explanations. The argument for non-representational explanation of pretence is not yet settled in this chapter, as this chapter does not engage with specific aspects of pretence at all. This chapter crucially sets the stage for these later arguments to have a fighting chance against representational explanations, it aims to bracket general worries with providing such alternative explanations. It also has a wider significance: with this chapter the reader finds out that while in some sense the topic of pretence is small and focused, considering it will involve some of the deepest and most difficult topics in philosophy, such as when is the use of mechanistic explanations or the causalconstitutive divide appropriate in philosophy. 2. Constitutive vs. explanatory role of mental representations Pretence can be conceived of in different ways; one may ask a constitutive question about what is the essence of pretence or an explanatory question about what mechanisms underlie it. It is important to keep these questions distinct. It will become apparent that many theorists who investigate pretence begin by assuming that pretence is mental-representational in nature. That leads them to justify using mental representational explanations for how pretence comes about. I will argue that this is not a valid justification, and that the questions of constitution and explanation can be kept separate. What follows is a discussion about the relationship between essences and mechanisms. Philosophical work usually aims to address the constitutive question (what is the essence of X), while cognitive science approaches (and some naturalistic approaches to philosophy) seek explanations of how it is that X is accomplished. Applied to pretence, contemporary theorists invoke representations of some sorts in the answers to both questions. Constitutive question about essences is asked to give an account of what is the nature of pretence. In the philosophical tradition, pretence is said to be a representational state of mind. Pretence has been traditionally defined as 3 requiring representational capacities, such as i.e. imaginations, symbolism, belief/desire-like states, concepts or double knowledge of what is real vs. not real. In a sense, the explanandum has been construed in representational terms. There are as many definitions of pretence as there are people who write about it, but mainly the most important ‘ingredients’ of pretence across the theories are said to be, in no particular order, possession of conceptual knowledge, having knowledge of what is real and not real, imagination, intentionality, being in a right mental state, having mental attitudes that are ‘belief-like’, shifting perspectives, mapping and following rules, symbolic thinking, and meta-communicating (Piaget, 1945/1962; Leslie, 1987; Harris and Kavanaugh, 1993; Lillard, 1993a; Mitchell, 1994a, 2002; Goldman, 1998, Nichols and Stich, 2000, 2003). Taking collectively from different papers, three common and important aspects of pretence are: (i) Double Knowledge (DKN): “The actor must entertain two representations of a single situation, one literal/veridical/real and the other nonliteral/distorted/pretend/imaginary, simultaneously, without confusion, and deliberately (Piaget 1945/1962; Leslie 1987; Harris and Kavanaugh 1993; Lillard 1993a; Mitchell 1994a, 2002)” (Russon in Mitchell (2002), p. 237). (ii) Imagination (IMG): “Pretence or make-believe is a mental activity involving imagination that is intentionally projected onto something (Goldman 1998, Lillard 1994)” (Mitchell 2002, p. 4); “All authors agree (…) that pretend play implies some displacement from the reality plane and thus meaning transformations relative to the meaning that the actions and the objects involved would have were they considered “literally” (Veneziano in Mitchell (2002), p. 60). (iii) Presence of mental plans or models (MPMs): “Internal plan recedes and guides the pretend action”; “The capacity to utilize such internal models of previous experience is considered to be the foundation of the capacity to engage in mental representation, and hence pretending” (McCune & Agayoff in Mitchell (2002), p. 47, 45).2 2 These have been argued by different theorists to be necessary for pretence. As can be noticed, these aspects happen to be ones that are often appropriated to the internal mental capacities of an agent; nonmental or external factors are scarcely mentioned, such as, for example, non-serious emotional tone or 4 The other question that can be asked about pretence is how pretending gets to be implemented. Here we will find those who think that mental representations of some form will figure in our best explanations of how pretending is accomplished. Most quoted and most elaborate current theories that are said to explain how it is possible to pretend are Leslie’s (1987) metarepresentational theory, various so-called ‘behaviourist’ theories (Stich and Nichols 2000, 2003; Perner, 1991; Harris, 1994), and intentionalist theory (Rakoczy et al., 2005). These have been discussed in detail the previous chapter. To re-count the relevant part, what the theories propose to do is, inter alia, to provide answers to variations of these three ‘how’ questions: (i) How can children hold a distinction between what is real and what is not real, how they do not get confused by pretence, or how can children use one object “as if” it is another one without losing track of the reality? (ii) How do children remain or persist in a pretence role, follow-up on their pretence, or step outside pretence when convenient? (iii) How do children choose one pretence act over another, produce contextappropriate play or play in a structured way? To answer these questions, pretence theorists invoke mechanisms that involve mental representations. We can characterize the three functions that mental representations are supposed to serve as (i) function of discriminating, (ii) function of transforming, (iii) function of guiding. The theorists then invoke these exemplar explanatory mechanisms to explain how the following work: (i) a mechanism of ‘quarantining’ (Leslie, 1987) or ‘flagging’ (Harris, 1999) to discriminate what is real from not real, or the ‘true’ from ‘pretend’ meanings; (ii) a mechanism of ‘decentring’ to transform one thing into another, which underlies the capacity to ‘imaginatively see as’ (Currie, 2004, 2006); or (iii) a guiding mechanism that manipulates propositions or forward models and images to guide one in ‘accurately’ acting the pretence scenario out (Nichols and Stich, 2001; van Leeuwen, 2009). By now, it should be apparent that mental representations get to come into the story through both doors: as essences of pretence, and as part of its supporting mechanisms. The three questions that aim to explain how pretence works (how agents involvement of bodily activity. These two features of pretence Lillard considers as ‘additional’ to pretence (in Lewis & Mitchell 1994, p. 214). 5 discriminate, be transformed and be guided) mirror the constitutive assumptions about the nature of pretence – that it involves the mental states of Double Knowledge (DK), Imagination (IMG), and being guided by Mental Plans and Models (MPMs). Often, pretence and mental representation are put on par; pretending is to be mentally representing, as, for example, the last quote of McCune and Agayoff shows (“…capacity to engage in mental representations, and hence pretending” (idem, emphasis added). Pretending is also often seen as a mental representational ‘state of mind’ or ‘attitude’. We could ask, are the commitments of these theorists ad hoc? When asking whether pretending requires representing, are they pre-supposing what they are trying to explain? While that may be likely, I will not engage in whether the reasons for believing that pretence is essentially mental representational are valid, as that is orthogonal to my main point. What is important is what follows from taking such a stance. Namely, there seems to be a consensus that some form of mentally representing must be involved in explaining pretence, because pretence is an essentially mental representational state to begin with. This is a line of thought that can be discredited, as even if some phenomenon is representational, it need not use representations in explanations of how the phenomenon arises. It is all too easy to claim that a child has Double Knowledge because of the quarantining mechanism in play; a child can has Imagination because of the decentring mechanism in play, etc. However, this argumentation would be fallacious. Consider linguistic practice as an example. Language can be said to be a representational activity, in essence. To use language, in all its forms, is at root to represent how things are, one way or another. However, that does not necessitate invoking a mental representational mechanism. Many have argued already that we need not have mental representations to have representations in the environment (Dennett, Price, etc.). So the linguistic capability can be explained without invoking mental representations, such as through socio-cultural, practical accounts (see, for example, Hutto’s (2008) Narrative Practice Hypothesis). Pretence may be characterized as a representational activity tout court just as language is, which means that by its nature pretence represents in a linguistic sense. We see such characterization plenty in the literature, when pretending is taken on par to be symbolizing (e.g., Lillard 1994). Pretence can be thus thought of as essentially a representational phenomenon: in pretence, something stands for or symbolizes 6 something else, so in that respect it is representing. However, we can separate the question of whether one needs to invoke mental representations to implement pretence, even if seen as a representational activity. The point is simply that even if when we’re pretending we’re representing, that does not mean that what enables us to pretend (or represent) involves further (mini-)representations. That would be analogous to saying that a house has to be made out of little houses; it is simply an unnecessary move.3 Just as with language, it is logically possible for pretence to be constitutively representational, while mental representations need not be invoked to explain it. One can think about pretence as essentially representational (or symbolic) without having to invoke symbol-swapping on a sub-personal level, which is one of the things invoking mental representational mechanisms is supposed to achieve (see, for example, Leslie). Thus, while ‘being representational proper’ could describe the system as a whole, that is irrespective of whether mental representations are to be used to describe its implementation or not. Interestingly, this does not seem to hold other-way around. That is, while mental representations ‘constituting essence’ need not require invoking mental representational ‘implementation mechanism’, claiming that implementation mechanism is mentally representational seems to say something about the constitutive essence of the phenomenon at hand, specifically, that it is mentally representational too. This holds for identity theory and some forms of functionalism. So are we justified in assuming that mental representations invoked in implementing pretence are not at the same time constitutively essential of pretence after all? Answering that question lies heavily on what notion of explanation one has in mind. This will be the topic of the next subsection. For now, it is enough to accept that, irrespective of whether we buy into the view that mental representations both constitute and implement mental phenomena or not, we can accept that mental representations play different functions as explananda and as explanans, even though cognitivist theories do not draw a sharp divide between them (Ramsey, 2007).4 It is my aim to dissociate the explananda from the explanans, and focus solely on the explanans. Thus, to argue for non-mental representational 3 I’d like to thank Sam Coleman for this analogy. As Ramsey claims, “the distinction between theories that posit representations and theories that try to explain representation is not as sharp as one might assume. A large number of cognitive models – indeed, perhaps the majority – do a little of both” (p. 36). 4 7 explanations of pretence, I will not provide arguments on what is the nature of pretence.5 What I am interested in is an explanation that does not need to make use of mental representations. The alternative I suggest comes from enactivism. It is a different methodological tradition to the cognitivist one. Although I will, for the rest of this dissertation, focus on the explanans (what mechanisms are in place that explain pretending), I also point out that with enactivism, it is possible to re-characterize the explananda. While logically the re-characterization should not be required, doing so upon occasion makes it easier to see how a non-representational explanation would work. Thus, based upon the above considerations, I will assume that there is a difference between essences and mechanisms, in this thesis I will be solely concerned with how mental representations enter into explanations of pretence, not whether pretence is essentially mentally representational. 3. Different approaches to explaining cognitive phenomena There are different approaches one can take to investigate cognition in philosophy of mind. That applies to investigating pretence as well. As I mentioned before, on the one hand, one can seek to know what pretence is and focus on its nature. That is a question of determining its conceptual essence. (This is the question I will not engage in any more in this chapter.) On the other hand, one can ask how we explain how pretending comes about. This section deals with looking more carefully at the explanations, about which another distinction can be made. One kind of explanation asks for implementation mechanisms of the explanandum, examples of which were mentioned in the section above. These are synchronic mechanisms that explain how pretending gets implemented at a time. They are also constitutive of what they explain; they form what metaphysically suffices for pretence, or, in other words, how it is realized (they need not, however, be essential, as there is room for multiple-realizabiliy). Another kind of explanation asks for external factors to the explanandum. These form causal explanations, which are not constitutive of the explanandum. They are diachronic mechanisms that aim to 5 For example, it could be that pretence does not involve mental representations in its explanatory role even if it is representational in essence, just as it could be that pretence still requires invoking mental representations to explain it, even if we conceive of pretence as a form of behaviour, not a mental state. 8 explain how, for example, a particular act of pretence (token explanandum) came about. In what follows, I will clarify what kind of explanation functionalism and enactivism seek. What I will point to is that functionalism only considers implementation mechanisms, while in enactivism, the distinction between implementation and causal factors (or synchronic and diachronic explanans) gets blurred. The issue is complex, and I intend to untangle certain commitments that are often left untold when investigating the topic of (social) cognition in philosophical methodologies. In this thesis I will specifically target pretence, and the explanatory models and mechanisms of pretence offered by various philosophers. a) Functionalism While there are many notions of functionalism (Sellars, 1956; Lewis, 1966; McLaughlin, 2006), the uncontroversial claim made by most functionalists is that the synchronic mechanisms ultimately explain phenomena, by acting as implementation mechanisms, or realizers, of functional roles. The mechanism is constitutive of the essence, but it can only be sufficient and not necessary to it (the mechanism can change as the role can be multiply realised). Realizers are constitutive of the phenomenon; they are an integrated part of what they explain, as if they were its ‘ingredients’. Consider an analogy to a clock, functionally understood as a device for measuring time. Its ability to measure time is realized by the internal mechanism of the clock. Its mechanism is the integral part of the clock, making it a clock (or sufficing for it to be a clock functionally understood as time-measurer) and at the same time explaining how it works (in the sense of being how it works). Importantly, the claim is not one of causality, as causal factors lay ‘outside’ of what we are explaining (they explain diachronically), while realizers are integral parts of what they explain (they explain synchronically). 6 The mechanism 6 While buying into role-realizer distinction, and proposing existence of realizers in explaining pretence, does not necessarily commit one to functionalism, it is not clear how one could step away from functionalism here, as no other approach to philosophy of mind has used this terminology so far. It is also not helpful to learn that the notion of realization has been used in many different ways. Papineau says of realisation, “There is no agreed analysis of this notion, however: philosophers tend to use it more freely than they explain it” (Papineau 1993a: 24, cf. also Tye 1996: 40-42, Melynk 1997: 390–395). In a bid to redress this, he offers the following definition: “In order for a mental or other special type M to be realized by an instance of some physical type P, M needs to be a second-order property, the property of having some property which satisfies certain requirements R. And M will be realized by P in some individual X if and only if this instance of P satisfies requirements R. In such a case we can say X satisfies M in virtue of satisfying P” (Papineau 1993a: 25, emphases original). 9 doesn’t cause the clock to be, for it is its clock master who made the clock into existence. In a token standard clock, its mechanical mechanism is sufficient for the clock to tell time. However, it is not a necessary mechanism; we can imagine a digital clock, whose mechanism is built of completely different parts, sharing nothing of the original clock’s structure (e.g., projecting numbers instead of moving handles, and so on), which still realizes the clock’s function to measure time. In that respect clocks can be multiply realized. Functional mechanisms give us multiple ways to instantiate the explanandum through their multiple realizability. As Bickle (2013) explains, In the philosophy of mind, the multiple realizability thesis contends that a single mental kind (property, state, event) can be realized by many distinct physical kinds. A common example is pain. Many philosophers have asserted that a wide variety of physical properties, states, or events, sharing no features in common at that level of description, can all realize the same pain. (…) [For example], a given psychological kind (like pain) can be realized by many distinct physical kinds: brain states in the case of earthly mammals, electronic states in the case of properly programmed digital computers, green slime states in the case of extraterrestrials, and so on.7 How does this analogy to clock pertain to pretence? I will first explain how speaking of realizers of pretence would look like, after which I will suggest an alternative. We can think of pretence in a similar way to a clock. For example, of an instance of pretence we could say it exemplifies the property, or role, of Imagination, i.e., just is an episode of Imagination. That property would be realized by a mental representational mechanism of, for example, ‘decentring’ (Currie 2006). On this story, Imagination is a property of pretence, which is part of the functional role of pretence (transforming), and what realizes it is the decentring mechanism. From this description, it should be clear that when applied to pretence, proposing a mechanism of ‘decentring’ should be treated as constitutive of and sufficient for ‘Imagination’, though perhaps not necessary (unless the philosopher at hand buys into identity theory or claims that his mechanism is the actual, not just possible, explanatory mechanism of pretence). Nichols and Stich (2001), for example, who propose ‘decentring’, 7 Stanford Encyclopedia of Philosophy, entry on ‘Multiple Realizability’. Retrieved on 11.02.2015. 10 engage in philosophy of cognitive science. They claim not to be offering constitutive formats, but whether indeed they do so or not depends on this debate. I believe their theories do answer the constitutive questions about the nature of pretence, even though that is not what they aim at, as they offer a substantive mechanism of pretence.8 We seem to be discovering through scientific investigation what is the essence of pretence and what its roles are. Under the functionalist model, we often speak of mental states and their properties, such as the mental attitudes of belief having a functional role of tracking truth, or the mental attitude of desire having a functional role of motivating. Following that picture, pretence would have to be considered as a mental state (like a ‘mental representational state of mind’ or a ‘mental representational attitude’), with a unique functional role of knowing what is real and not real (DK), imagining counterfactual scenarios (IMG) and being guided to play in a specific way (MPM). Thus, what we gain from taking a functionalist approach is the following picture is the following. Knowing real from not real, imagining and being guided can be treated as aspects of the functional roles of pretence, both constitutive and essential to pretence. (Mental states like DK, IMG and MPMs embody these roles into properties of pretence.) Pretence mechanisms, like decentring, quarantining, etc., implement these roles of pretence; as realizers, they are also constitutive but not essential to pretence. However, there is a worry with taking the functionalist approach when studying cognitive phenomena. With regard to considering pretence, do we have reasons to think it is anything like clocks? Is it something that can have synchronic realizers, or can be broken down to constitutive subpart(s)? Can pretence be explained without taking under consideration, for example, the context in which it occurs? I would tend to say ‘not necessarily’ to all of these questions, but that leans heavily on what notion of explanation one has in mind. That will be engaged in below in subsection 4, but before that, I will put another approach to studying cognition on the 8 There is an ongoing debate about what is supposed to be the job of the philosopher. Some say that it is to answer constitutive questions about the nature of things (e.g., what is the nature, or essence, of cognition). These are the ‘traditional’ philosophers. However, we have also less orthodox philosophers, ones engaging in cognitive science. According to Barry Smith, philosophers play part in larger scientific enterprise and make interdisciplinary contributions (in Haug (ed.) 2013, p. 295). To take another example, David Papineau is interested in the actualities, not necessities (idem, p. 192).8 Thus, there are different ways in which philosophers are interested in the constitutive question. 11 table – enactivism. It is crucially different from the other two, and it is highly relevant to this thesis as it is the approach that promotes non-representationalism. b) Enactivism There are many variations of enactivism, the spectrum varying from biological enactivism that stresses autopoietic structuring of a living organism like a cell (Varela et al., 1991), to radical enactivism that stresses lack of mental, semantic contents in early forms of cognition (Hutto and Myin, 2013). However, there are some aspects of enactivism shared by all (or most) parties. The first is that enactivism is (or should be seen as) a non-representational account of cognition. Enactivism questions most commitments of cognitivists (and connectionists), against which it has been proposed. As Varela et al. (1991) claim, It questions the centrality of the notion that cognition is fundamentally representation. Behind this notion stand three fundamental assumptions. The first is that we inhabit a world with particular properties, such as length, color, movement, sound, etc. The second is that we pick up or recover these properties by internally representing them. The third is that there is a separate subjective “we” who does these things. (…) We propose as a name the term enactive to emphasize the growing conviction that cognition is not the representation of a pregiven world by a pregiven mind but is rather the enactment of a world and a mind on the basis of a history of the variety of actions that a being in the world performs (p. 9). The second is that enactivism is modeled on ecology, not computation: Ecological approach stresses the dynamic relation between the organism and its environment, whereas computational approach focuses on manipulation of symbols or representational mapping. Enactivism, thus, proposes the study of living organisms, and describes autopoietic systems in their environments. What is interesting is Varela at al.’s (1991) comparison of such autonomous systems from mechanistic systems. They write: Such autonomous systems stand in sharp contrast to systems whose coupling with the environment is specified through input/output relations. The digital computer is the most familiar example of this latter kind of system. Here the meaning of a 12 given keyboard sequence is always assigned by a designer. Living systems, however, are far from being in this category. Under very restricted circumstances can we speak as if we could specify the operation of a cell or an organism through input/output relations. In general, though, the meaning of this or that interaction for a living system is not prescribed form outside but is the result of the organization and history of the system itself (idem, p.157). The benefit of enactivism is that it considers the individual in the intersubjective context. What that means is that it takes under account the role of objects and others in shaping cognition. The outcome of taking an enactive approach to cognition is that it finds mentality in the interactions, not in encapsulated mental representations. The third aspect of enactivism is that it stresses the constitutive role of the body and environment in shaping the mind. That is, whatever pertains to mind is going to be necessarily involving the body of the organism and its world. What is important for our discussion is that on enactivist picture, the causalconstitutive divide falls apart. Enactivism talks about the holistic explanations, taking under account the organism’s biological set-up, history of interactions, social and culture niches, etc., and so there is no clear divide between what is the synchronic or diachronic explanation. For example, the history of interactions is both necessary to explain how a mental act came about now and it can be used to explain how it got to come about yesterday. Though there is no clear application of enactivism to pretence yet – as establishing this possibility is the aim of this thesis – we can already imagine that enactivists would claim that there is nothing ‘extra’ to the practices that enactivism explains that realizes them ‘from within’, as there is no internal-external divide to be made. Pretence would be necessarily embodied and situated in a particular context. There may be nothing extra to realize ‘pretence’ or its aspects (roles) like DK and IMG (non-representationally construed). Rather, it may be the history of interactions that shape relevant know-how, or sensorimotor and perceptual capacities that allow imaginative ‘seeing-as’, which are enough to explain pretence. What happens to the explanatory notion of enactivists that serves as an alternative to mental representations? While in the functionalist framework, we could speak of them as ‘realizers’ of pretence as they take on explaining the role of, e.g., 13 ‘imagining’ or ‘guiding’; in the enactivist framework, they should not be seen as realizers of pretence, because affordances cannot be seen as constitutive subparts of pretence as mental representations are; affordances are to be thought of as either relations between environment and animal (Chemero 2009) or relational properties of the environment relative to the animal (Gibson 1979, Noe 2001, Withagen et al. 2012), and as either of the two, they are prone to dynamical interactions. To summarise, in this section I have explained that the investigation of a cognitive concept can take three forms: investigating its conceptual essence (answering the question: what is it?), investigating implementation mechanism (how is it realized?) and investigating its external factors (how does it causally come about?). I have explained that on the functionalist picture, the first two are interlinked, and contrast the third: findings about the implementation mechanisms (constitutive realizers) can tell us something important about the essential make-up of what we investigate. However, on the enactivist picture, the first stands separate to the second and the third, which are, in turn, tightly linked; so tightly that it may not make sense to separate them. Taking this under consideration, how does enactivism propose alternative explanations to mental representational explanations of pretence? Are the alternatives to be considered as mechanisms at all? Whether a mechanistic explanation needs to be posited will be clarified upon discussion of the nature and success of mechanistic explanations, in the subchapter below. 4. The validity of mechanistic explanations To show how pretence can be understood without mental representations, we must understand how pretence is usually explained in the first place. It is usually explained though a cognitive mechanism. These mechanisms showed which aspects (‘roles’) of pretence needed explaining, out of which I have chosen three (knowing, imagining, guiding) to explain with a non-representational alternative. But what would a non-representational mechanism look like? Can we even have nonmechanistic explanations? These questions need to be at least raised (if not answered) before developing the positive proposals in order to clarify what exactly it is that the non-representational alternative provides. The types of explanations present in philosophy of mind have mainly been mechanistic. As Withagen et al. (2012) claim, “Ever since the mechanization of the 14 worldview (…), philosophers and later psychologists have generally tried to understand mind and behavior in mechanistic terms” (p. 250). Descartes (1641/2007) was a famous promoter of such methodology, but more recently, the method of breaking the phenomenon into smaller ‘component’ mechanisms that are ‘stupid’, and therefore, have lesser functions, has been discussed most famously by Dennett (1993). While there is no one accepted definition of a mechanistic explanation, Ramsey (2007) suggests the following: A mechanistic explanation accounts for some capacity or aspect of the mind by showing how it comes about through (or is realized by) its structures, states and operations that could be implemented in physical process (p. 25). As it can be seen, treating aspects of pretence as roles that have realizers (decentring, quarantining, etc.) fits with the mechanistic view. 9 These aspects are something into which we can ‘break down’ the phenomenon in order to explain its smaller component parts. Why scientists and philosophers want to introduce mechanistic explanations is to break the investigated phenomenon down into a smaller set of components, so that they are easier to interpret and can be put back together to allow making predictions about when the phenomenon would emerge. For example, knowing what is required of pretence in terms of components would give us an idea of what parts we would need to ‘collect’ so that when we put them together, pretence would emerge. However, such mechanistic explanations do not seem to go in pair with enactivist framework, the reasons why will be clarified after I introduce some examples of mechanistic and non-mechanistic approaches to cognitive phenomena. Marr and Poggio’s (1976) paper is famous for introducing a differentiation between levels of description; it was applied to understanding perception. They approached perception from a computational theory approach, threating is as an information processing system. As such, they distinguished three levels of abstraction in analyzing perception (in Dawson, 1998): 1. Physical level: How the system is physically realized (neuronal structures), 2. Algorithmic/representational level: How does the system do what it does, 9 Interestingly, Ramsey himself does not see how a properly mechanistic framework of a cognitive theory would include roles such as ‘informing’, ‘denoting’ or ‘standing in for something else’ that mental representations are supposed to play (2007, p. 25). 15 3. Computational level: What does the system do and why it does these things. With Marr’s consequent paper (1982), the three levels of explanation have become popularized. In reversed order, Stich and Nichols (2003, p. 10) explain these levels in the following way: Marr’s highest level, which he perhaps misleadingly called the computational level, presents an analysis of the cognitive capacity to be explained. The analysis given an explicit account of what the system underlying the capacity does and decomposes the system into a set of simpler subsystems. (…) At Marr’s second level, which he calls the algorithmic level, the theory provides formal procedures or algorithms for accomplishing each of the tasks into which the initial target task or capacity has been decomposed. The third, or physical implementation, level explains how the algorithms are actually executed by some real physical system – typically some part if he brain. On Marr’s view, the three levels are largely independent of one another. While it hasn’t been explicitly proposed, pretence could be re-described also in the following way: 1. Physics: what is happening at the neuronal level, 2. Functional Mechanisms: How does the system do it? (One possibility: mental representational models) 3. Action: What does the system do? (Pretending, could be understood as representing proper) As the first domain clearly belongs to physics or neuroscience, not philosophy, it will not be discussed further. Interestingly, philosophers do take on themselves to explain the second level by providing explanatory models. This level is about asking the question: how does the system get to do what it is doing? Applied to pretence, it can be asked what structures or capacities need to be in place for an agent to pretend that a banana is a phone? Here, mental representations are brought in in forms of (often modular) mechanisms to explain what enables pretending to take place. At the third level, we may ask, what is the child doing when she is pretending that a banana is a phone? One answer might be that the child is representing the banana as a phone. That notion of ‘representation’ should be understood as ‘thinks of’ or ‘takes one to be 16 another’, a kind of a symbolic thought. This is the level of description that many target when they defend the idea that pretence involves representing proper. For the purposes of focusing on explanatory mechanisms, how we describe pretending at the highest level will also be bracketed. Nichols and Stich target the ‘top’ (computational) level when proposing their account of mindreading, of which pretending is an example. They propose a modular mechanism to play a role at the second level. They write: The account of mindreading we propose in this book can be viewed as a theory at Marr’s highest [a.k.a., computational] level of analysis or, what amounts to much the same thing, as the first level of a homuncular functionalist theory of mindreading. Our aim is to characterize the complex and variegated skills that constitute our mindreading capacity and begin the job of explaining how these skills are accomplished by positing a cluster of mental mechanisms that interact with one another… (2003, p. 10-11). They further claim that Most of the competing accounts of mindreading that have appeared in recent years (along with many cognitive theories aimed at explaining other complex cognitive capacities) are best viewed as having the same goals as our theory. They try to explain mindreading (or some other complex cognitive capacity) by positing functionally characterized underlying mechanisms with capacities that are simpler than the capacity they are trying to explain. (idem, p. 11) Thus, when following Marr’s methodology, phenomena like ‘mindreading’ are broken down to simpler mechanisms in need of further explanation. Interestingly, Marr himself made clear that his levels of classification are to be treated as independent of one another. What should follow is that the algorithmic/representational level should not in any way affect (whether strengthen or weaken) the computational level; these can and should be considered separately from 17 each other. That is why logically, the structure of the mechanism of the middle level should not make a difference to the structure of its highest level.10 Interestingly, this is the one key point that Nichols and Stich decide not to follow Marr on. They admit to one major change in their analysis after appropriating Marr’s distinctions: Though it may be helpful to think of our account of mindreading as a theory at Marr’s ‘computational’ level of analysis, there is one important issue on which we differ from Marr. On Marr’s view, each level of analysis is largely independent of the level below. On our view, by contrast, findings about the structure and functioning of the brain can, and ultimately will, impose strong constraints on theories of mindreading of the sort we will be offering (idem, p. 11). Nichols and Stich are thinking of the lowest (physical) level affecting the algorithmic level, and find that not worrisome due to lack of sufficient findings about the brain to make any changes on the next explanatory level. My attention is drawn to a different connection: if the levels are not independent of each other, then the computational level is to be supported by what is broken down to an algorithmic level. Their picture is of homuncular mechanisms that entail each other, rather than three separate levels of analysis of the same phenomenon. The reference to the scientific approach of Marr is tainted. Unfortunately, this entailment relation is also endorsed in different fields. Other authors (Iacoboni, 2006; Gallese et al. 2004) used a similar distinction to speak about empathy and social cognition. For example, Gallese et al. (2004) distinguished these thee levels underlying understanding of social cognition: 1. Physics: Firing of mirror neurons in the brain 2. Syntax: The functional model of simulation (simulation theory) 3. Semantics: Conscious experience, phenomenology of empathy. As Gallese and his colleagues are, first and foremost, neuroscientists, they mainly focus on explicating the neuronal mechanisms underlying empathy at the physical level. However, they also claim that the neuronal mirroring activity is the 10 This resembles the enactivists’ divide between mechanisms and essences, but not one of functionalists, who take the realizer mechanisms to be telling of the essential make-up and roles of their explananda. 18 physical basis of our action-simulation capacity, which seems to be positing explanatory modules on the second level. That mechanism of simulation further explains how it is that humans feel empathy (third level). Thus, while providing a hierarchical, mechanistic decomposition of phenomena is a common scientific practice, that traditional approach of looking for underlying mechanisms of phenomena may not be best applicable to complex social and mental phenomena; it even results in contradictions.11 Explanations of cognitive phenomena perhaps should not resort to looking for simple mechanistic explanations. Two alternatives can be thus proposed. The first is to get rid of the mechanistic explanation altogether. The second is to embrace that the explanation can be considered as a ‘wide-‘ or ‘extended mechanism’ that expands our notion of mechanism to include social structures and the environment. I will now discuss both options. On the one hand, speaking of mechanisms may be problematic for enactivists, and many would take a ‘no mechanism’ view to providing explanations. One reason is that the idea of mechanistic explanations is that they just have to be general – this is part of what it means to explain something. To apply this thought to pretence, to explain types of pretence phenomena would be to looking for mechanisms that instantiate pretence. But perhaps there is no unique phenomenon of pretending that would have such a general mechanism. That is because often there is no one accepted definition of such concepts, or a unitary account of what can be covered by such concepts, and without such information, it is difficult to establish whether what we are trying to explain even is a kind of a thing that can be broken down to smaller mechanisms. Consider Hutto’s opening lines to “What is Consciousness?” (2014): There is no utterly clean, clear and neutral account of what exactly is covered by the concept of consciousness. The situation reflects, and is exacerbated by, the fact that we speak of consciousness in many different ways in ordinary parlance. A consequence of our multifacious uses of the concept is that is has impossible to define its essential characteristics through conceptual analysis. We have nothing approaching a descriptively adequate philosophical consensus of what lies at the core of all and every form of consciousness in terms of necessary and sufficient 11 See Rucinska (2015) on Fuchs 19 conditions that would be accepted by all interested parties. (…) Recognizing that attempts to provide a philosophically robust definition of consciousness are likely forlorn, a standard tactic for isolating core features of consciousness is to provide clear-cut exemplars as specimens. While Hutto speaks of consciousness (not pretence), and the inability to agree on one definition of consciousness (not account), his words speak to pretence and its available accounts as well. What follows is that it might be a daunting task to look for a mechanism of pretence, one that would unitarily underlie any pretence act. For aside no agreed definition of pretence or its account, there is no one kind of pretending either. As was mentioned in Chapter I, pretence formulates itself in various ways, from object-substitution play, to role-play, to full on acting and deception. It is difficult enough to support the idea of a set of ‘components’ of pretence that would be entailed by all these forms of pretence, let alone to propose a general mechanism that specifies how those components interact with each other to enable pretence. Moreover, mechanistic explanations do not seem to go with enactivism, as aside assuming that the phenomenon can be broken down to smaller components, it assumes that we can posit general rules that represent how things should happen. However, giving rules of ‘pretence’ would be to de-contextualize it, so that the component parts of a functioning ‘mechanism’ would always hold for any pretence case. The worry is that maybe we cannot speak of a decontextualized mechanism of an activity like ‘pretence’ to start with. Consider Withagen et al.’s (2012) explanation of why Gibson’s ecological psychology tried to do without mechanisms: Gibson’s (1966,1979/1986) ecological psychology can be understood as a critique on mechanistic metaphors that dominate thinking in psychology since the 17th century (Reed, 1988a, 1996a; Withagen & Michaels, 2005). (…) In the beginning of the 20th century, Holt (1914/1973), Gibson’s graduate school mentor, claimed that “psychology is at the present moment addicted to the bead theory” (p. 160) – it tried to understand behavior in terms of a chain of causes and effects. Following his mentor, Gibson (1966, 1979/1986) took aim at the mechanistic conceptions of perception and action. He argued that animals should not be conceived of as machines, the responses of which are caused by stimuli from the environment. In addition, he claimed that the mechanistic conception of the environment as matter in motion is inappropriate to understand animal 20 behavior. In Gibson’s view, the animal’s environment consists of action possibilities, which he termed affordances. For instance, in the human environment, a floor affords walking upon, a cup affords grasping, water affords drinking, and so on. With this new conception of the environment, Gibson put agency back on the agenda and thereby overturned the mechanistic framework that underlies many psychological approaches. After all, if the environment consists of opportunities for action that do not cause behavior but simply make it possible, animals appear as being autonomous, making their way in the world. They are not mere puppets pushed by the environment like machines; rather, animals have agency (p. 250). On the other hand, perhaps mechanisms can be appropriated by enactivists, and the ‘extensive mechanism’ view could be endorsed. The ‘extensive mechanism’ view would be one answer to the worry of decontextualization. Carlos Zednik (2011) argues that we can appropriate a dynamical explanation to explain complex and distributed mechanisms. He claims: The theoretical frameworks of computationalism and connectionism are often construed as a search for cognitive mechanisms, the specific structures and processes from which cognitive phenomena arise. In contrast, the framework of dynamicism is generally understood to be a search for principles or laws — mathematical regularities that govern the way cognitive phenomena unfold over time. (…) I argue that dynamical explanations may be uniquely able to describe mechanisms whose components are engaged in complex relationships of continuous reciprocal causation (p. 238-239). One example how a wide, dynamic type of explanation is how Varela at al. (1991) explain colour experience. In “The Embodied Mind”, Varela et al. discuss how colour experience comes about. To understand colour experience, they claim that we need to understand factors that simultaneously shape the experience, such as appearance of the object (structured by basic colours and dimensions of hue, saturation and brightness), illumination, retinal structures, neuronal structures (structures of visual pathways that participate in the perception of colour), relationship between colour and motion, and other sensory modalities. They claim that “Our coloured world is brought forth by complex processes of structural 21 coupling” (164), and they demonstrated that “colour as an attribute is intimately involved with other attributes of our perceived world” (165). They conclude with: “Our examination so far shows that we will not be able to explain colour if we seek to locate it in a world independent of our perceptual capacities. Instead, we must locate colour in the perceived or experiential world that is brought forth from our history of structural coupling.” (165). According to them, colours belong to shared biological and cultural world. This is in line with their understanding of enactivism, which, according to them, merges realism, idealism and non-representationalism. They claim: Our discussion of colour suggests a middle way between these two (…) extremes [realism and idealism]. We have seen that colours are not ‘out there’ independent of our perceptual and cognitive capacities. We have also seen that they are not “in here” independent of our surrounding biological and cultural world. (…) These two extremes both take representation as their central notion: in the first case representation is used to recover what is outer; in the second case it is used to project what is inner. Our intention is to bypass entirely this logical geography of inner versus outer by studying cognition not as recovery or projection but as embodied action” (172). It is important to focus on what it is that Varela et al. are claiming to be explaining. It could be said that they explain what colour experience is. According to them, (…) the overall concern of the enactive approach to perception is not to determine how some perceiver-independent world is to be recovered; it is, rather, to determine the common principles or lawful linkages between sensory and motor systems that explain how action can be perceptually guided in a perceiverdependent world (idem, p. 173). Varela et al.’s example shows that it is possible to provide something more than a simple mechanistic explanation of a cognitive action. They do seem to endorse looking for ‘common principles’ or ‘lawful linkages’ (that belong to providing a mechanistic explanation) that can be generalized to all perception, but they look also at the relation of the sensory-motor systems to the external world, one that is perceiver- 22 dependent. It could be said that they have opened up their notion of a ‘mechanism’ to include social and cultural structures. It would be a type of an ‘interactive’, dynamical mechanism, one very much in line with Zednik’s proposal. There are worries to be found with depleting the ‘wide mechanism’ view. One worry is, how could social context be appropriated to anything like a ‘mechanism’ of pretence? How can such mechanism be studied or implemented? While it is promising that further research can be done to make use of the enactive framework in, for example, an empirical study (which is what an ‘enactive mechanism’ would promise), it has to be carefully considered whether we are speaking of anything like a mechanism at all, let alone whether we are speaking of it consistently. In his review of a book “Analytical Sociology and Social Mechanisms”, Bert Leuridan (2011) neatly raises the worry with applying the talk of mechanisms to concepts of social nature. He explains: Whenever we start explaining ‘why’ something happens, beyond mere description, we are necessarily led to introduce some type of causal linkage of elements that in turn raises the question of mechanism. (…) Emphasis on the notion of mechanism corresponds to an evaluation of the proper role of causal linkages in the social sciences. Yet there is no consensus on the notion of ‘social mechanism’ in analytical sociology. (…) In Analytical Sociology and Social Mechanisms, one hardly finds explicit definitions of ‘mechanism’, while the concept is used implicitly in many divergent ways. As this is a big discussion, one that is beyond the scope of this research, perhaps I can remain neutral on whether the type of explanation I will be providing is a mechanical explanation. It may be wise to follow the steps of Tad Zawidzki (2013), who at the introduction to his account of mindshaping (importantly, a counter view to mindreading proposals present in the literature), says the following: (The) view I defend is entirely neutral on the question of mechanisms. The competencies I describe are not implemented by innate, special-purpose symbolic modules, or by general-purpose, experience-driven neural networks, or by some hybrid of the two, or by some other alternative” (2013, p. xiv). 23 If these are the options of how to interpret what mechanisms are – either taking the form of innate symbolic modules, or neural networks – then my alternative will definitely not propose mechanisms per se. Rather, what I will propose are, to put most neutrally, factors (non-representational factors) that play a role in making pretence happen. Taking this approach may not require taking a stance on whether the explanation is a ‘no mechanism’ or ‘wide mechanism’ explanation. However, should a stance be taken, for the purposes of this particular thesis, I do lean towards a ‘no mechanism’ view. That is mainly due to the description of mechanism that Ramsey proposes, which is that mechanisms are by nature representational: … (Try) to imagine what a non-representational account of some cognitive capacity or process might look like. Such a thing should be possible, even if you regard a non-representational account as implausible. Presumably, at the very least, it would need to propose some sort of internal processing architecture that gives rise to the capacity in question. The account would perhaps invoke purely mechanical operations that, like most mechanical processes, require internal states or devices that in their proper functioning go into particular states when the system is presented with specific sorts of input. But now notice that in the current climate, such an account would turn out to be a representational theory after all. If it proposes particular internal states that are responses to particular inputs, then, given one popular conception of representation, these would qualify as representing those inputs (2007, p. 3) As my proposal is to provide the possibility of non-representational pretence, I will cautiously accept Ramsey’s premise, even if it could be defended – even if mechanisms are not by nature representational. There is, however, no compelling argument for that option; in fact, even Zednik accepts that dynamical explanations are “reductive explanations (…) and may be amenable to representation hunting” (p. 239). That is why the non-representational factors that I will propose in explaining pretence can be, for present purposes, best conceived of as non-mechanistic explanations. 5. Interlocutor’s objections and enactivist responses 24 This is where a cognitivist interlocutor may raise an objection. Can we have non-mechanistic explanations that serve as explanations tout court? Are these to be counted as ‘explanations’ at all? Consider the concept of ‘affordances’ that was mentioned in Gibson’s quote, which will be a prominent non-representational explanatory notion I will use throughout the thesis: does invoking affordances provide a genuine alternative explanation, or are they just re-descriptions? The cognitivist may worry that with introducing such notions, I have not engaged in an explanatory work, but engaged in doing something else entirely. This is not an unjustified worry. Without structuring the explanations into mechanisms, I could be accused of not providing an explanation. I could also be accused of changing the topic to conceptual analysis. I will presently discuss these two objections, and propose rebuttals. The responses will also serve as summaries of main points made in the above sections. To begin to answer the cognitivists’ worries, and to set the stage for rebuttals, it must be mentioned that in the nature of what we are explaining, it is difficult to establish up front what kind of explanation would satisfy us. It is important to remember that there are at least two types of philosophical investigation. One equates explanations with saying something about the constitutive nature of what is explained; other does not. How constitutive questions about a notion X are answered could, in turn, be derived from pragmatic use of “X”, or from analysing the concept X. The role of X is hard to determine; often it is assumed. Take the concept ‘thinking’ as an example. Philosophers like Peter Hacker and Jason Stanley (2011) think that we first have to find out what ‘thinking’ is, by seeing how we use the term ‘thinking’ in our practice (Hacker prefers good old fashioned ordinary language philosophy for this task, and Stanley linguistic analysis). In the second step, they say, neuroscience can find out about the (actually boring) mechanisms, which realize ‘thinking’.12 For example, in functionalism, realizers serve as explanations of the roles. Then what realizes X can be mechanisms; not just physical mechanisms are used as explanations, but also theorists posit symbolic modules (see Leslie 1987, Nichols and Stich 2000, 2003). But the second type of investigation does not make that distinction. In asking about the phenomenon X, enactivists want to understand how it comes about. There is 12 It is not always the case that neuroscience or cognitive science plays that role; arguably it is a philosopher’s task. 25 no constitutive nature of X on that picture. Consider again the explanation of how colour experience comes about by Varela at al., which includes the description of the physiognomy of the human eye, the influence of the colour concepts of our culture on perception, and so on. It could be said that they holistically explain how it is that we get colour experience. Now, is there something else, the constitutive nature of colour experience on top of that? It is likely that they would say no; there just is the phenomenon of colour experience, and Varela et al. explain how it works (how it came about and persists at the time). 13 That must serve as explanation on their account. At this point a cognitivist interlocutor may pose the following, first challenge: the enactivist is not providing an explanation at all. That is because cognitivists are interested in the synchronic nature of the explanation, looking at the types of cognitive mechanisms that are at work at the very moment one is, e.g., pretending. Cognitivists insist that to provide an explanation is to provide an answer to a synchronic question - what makes a specific behaviour a ‘pretence’ behaviour at the moment we are engaging in it - as opposed to a diachronic question, which would entail some sort of a historical perspective. As the enactivist is not just after the causal antecedents of pretence, what is left is narrowing the target down to what synchronously (or at a particular given time) suffices for a pretence act. Realisation is a natural way of describing this relation. The objection would be that if one is not speaking about synchronic realization, or diachronic causal antecedents, then one is not providing a genuine alternative explanation to mental representational explanations of pretence. Hence, the working concepts ‘affordances’ cannot play the alternative explanatory role as they do not fit these explanatory roles to begin with. However, enactivists could argue that it is a genuine explanation, just one that does not fit with the cognitivist requirements. Consider an analogy: A policeman pulls over an intoxicated woman for drunk-driving and asks her: “When did you start drinking?”, to which she replies: “When I was about 12 years old”. This joke of an answer suggests something important: both this answer and one that would specify the hour of the day (ultimately what the policeman was asking for) are acceptable answers; they do answer the same question, and explain how the woman got to be drunk-driving, in a sense. 13 Thanks to Martin Weichof for his helpful commentaries on the topic. 26 Enactivists negate that synchronic explanations can be cleanly divided from the diachronic ones. They use 'history of past interactions’ as an explanation for why we act the way we do presently. Thus, the answer to the cognitivist interlocutor might be to argue that the enactivist explanation is a genuine explanation after all. ‘Affordances’ enter as explanations under a different methodology. They still clarify how pretence works, but the divide between synchronic and dialogical explanations gets lost. It is conceivable, thus, that one can provide a genuine alternative, as long as one is not playing by the same rules. Whether they are going to form aspects of a ‘wide-mechanism’ or form a ‘no-mechanism’ explanation is beyond the scope of this paper, though as argued in subsection 4, the ‘no-mechanism view’ may be the safer option. Thus, the alternative may not form an alternative mechanism, but should be counted as an alternative non the less. Speaking of realizers of pretence would stop making sense on that account; there would be no independent mental state of ‘pretence’ outside the practice of pretending; that would just be one and the same thing.14 The second challenge of the cognitivist interlocutor may be that the enactivists have changed the topic. Again, it is a valid point. Another option to not providing explanations of general phenomena is to speak of token phenomena, understanding that we cannot decontextualize them or give any golden rules. As mentioned earlier, there may not be, after all, a single coherent ‘pretence’ phenomenon. However, that point will always be ad hoc, unless one provides a working definition of ‘pretence’. Providing a definition, in turn, is engaging in a conceptual analysis, which is not what enactivism should approach if it is to provide alternative explanatory story. A possible response of an enactivist, then, could be to bite the bullet and accept that the explanation is not of the same kind, and indeed perhaps enactivism has changed the explanandum, but that it is a change for the better none the less, because explaining pretence by mechanistic explanation fails by its own standard (as proposed by Ramsey). However, importantly, the enactivist should be aware that his/her task is not to be engaging in philosophy of language, and what follows, conceptual re14 The enactivist would say that we cannot know in advance what the roles and realizers of pretence are, and assuming them is making a mistake, because we're then looking for some 'universal principles' of, e.g., pretence, and those would have to be decontextualized from the actual practices and the varieties they bring. If the practice of pretending is always contextualized, then we cannot presuppose roles (and realizers) of the pretence, and so the functions and the mechanisms these functions are supposed to play. 27 characterizations. I will not provide an explanation of what makes one behaviour count as pretence behaviour. One doesn’t need enactivism to answer the question what counts as what. Ordinary Language Philosophy provides better answers to “what counts as what within our linguistic practice” questions. The nature of the enactivists’ job is to discuss mental phenomena like pretence in the domain of philosophy of mind, in particular by focusing on alternative explanations of pretence, not on how the word ‘pretence’ is being used in a linguistic practice. There may be just the pragmatic practice of using “X” (‘pretends’) in different contexts, and the phenomena of X (pretence). But these things can come completely apart, because our ordinary language is not made for describing phenomena, but for pragmatic purposes (e.g., ‘thinking’ and saying things like “think harder!”). So there could be the real phenomena of pretending on the one hand, and the practice of using “pretence”, “pretends”, etc. on the other hand. That second analysis I have to bracket, and for the argument sake assume that ordinary language does a fine job in defining the boundaries of pretence. Even if that is not the case, this is not a job of an enactivist looking at explaining non-representationally how pretence comes about, even though on the enactivist picture, language games and cultural descriptions could come in as part of the wide, cultural explanations. 6. Conclusion: Working assumptions and strategy Does all this matter in finding non-representational alternatives to explanations of pretence? Yes, as it is necessary to clarify upfront what kinds of explanations I will be offering, and bracket any confusion of whether I am providing an explanation. Considering the nature of the explanation is important as it shows the possibility of interpreting my positive proposal (non-mental representational explanations) in subsequent chapters as genuine explanations of pretence. It is also crucial for getting clear on the working assumptions and the strategy appropriated for the rest of the thesis. For the purposes of this thesis, I will accept that some form of knowing, imagining, or guiding is involved in pretending, in order to discriminate genuine pretence acts from similar but not genuine acts of pretence. These form ‘roles’ on the functionalist accounts, but should not be thought of as such by the enactivists (as they do no not propose realizers of such roles). For example, I accept that some kind of 28 knowledge of what is going on is present (and some kind of understanding of what one is doing) – but whether a Double Knowledge state of mind is present is not to be taken for granted, and whether it requires a mental representational mechanism of ‘quarantining’ or ‘flagging’ will be put to question, etc. I will also re-describe each and every one of the aspects. To be clear, engaging in such re-description them may not be necessary, as it is possible to endorse mental representational aspects of pretence (on one level of description) and endorse their mechanisms as non-mental representational (on another level of description). That only follows if we accept that mechanisms do not constitute their essences. But in case mechanisms do constitute their essences, it is useful for clarification purposes to show that the ‘essences’ of pretence can also be characterized non-mentally representationally. While it is not strictly necessary, it will be easier to see how securing the possibility of non-representational explanans (the goal of this thesis) can get cashed out in explaining how pretence works when I show that pretence need not be necessarily seen as intrinsically mental representational to begin with. Thus, I will target the constitutive commitments to show that they are not necessary or even the only way to think about pretence. What follows is that explanatory needs will change, and a logical space is opened where such questions of explanation can be answered in a different way. 29