Abstract
What mental states guide the execution of our actions? It is generally agreed that the execution of an action is guided by the relevant knowledge state concerning how to perform that action. However, not all agree on which mental states underlie such a knowledge. Some suggest that knowing how to perform an action has mainly to do with the propositional representation about the way to execute that action. Those opposing this view stress the role of the motor, non-propositional representation as the mental state responsible for action performance. The aim of this article is to overcome this dichotomy by showing that an explanation of the cognitive processes underlying knowing how to perform an action needs both propositional and motor states. We defend this view by providing an account of the way in which our propositional knowledge about an action is constituted by the motor representation that guides the execution of that action.
What is the relationship between knowledge and action? It is a common assumption that, if we are able to successfully perform an action, it is because we know the way to perform that action. However, our knowledge of the way to perform an action does not always and necessarily lead us to be able to perform that action. For example, we may know that w is the way to perform the action F and, nonetheless, we may be unable to trigger the proper motor command needed to execute F in the way w. Circumstances like this raise an intriguing question: why, even if we know the way to perform an action, may we be unable to perform that action that way? In other words, what is needed, in addition to the knowledge that w is the way to perform F, in order to be capable of forming the adequate intention needed to execute F in the way w? In this article, we answer this question by arguing that, in order for our knowledge of the way to perform an action to be the guide of the successful execution of that action, this knowledge must be constituted by the relative motor representation of that action. In order to clarify our claim, we must proceed slowly. Let’s start with a case: Suppose Ariel knows that w is the way to successfully execute the action F. Problem: does Ariel know how to execute F in the way w?
To address this issue, 1 a preliminary distinction should be made. One may conceive the knowledge of the way to perform an action as an epistemic condition regarding the propositional content of a relevant belief: Ariel knows how to perform F in the way w, in the epistemic sense, if Ariel has a true and justified belief about the way w to perform F (Carter & Poston, 2018). One may also conceive the knowledge of the way to perform an action in a practical manner, that is, as an ability to perform that action in that way: Ariel knows how to execute F, in a practical manner, if she has the ability to execute F in the way w.
Whereas it is obvious that knowing that w is a way to F is enough for knowing how to F according to an epistemic sense, at least if we adopt the standard definition of knowledge, it is not so obvious whether it suffices to know how to F according to a practical manner. There is some consensus on the assumption that to know how to perform an action in an epistemic sense, a (true and justified) belief about the way to perform that action, expressed in a propositional format, is enough. 2 However, not all agree that knowing how to perform an action in a practical manner only has to do with contents expressed in a propositional format. Notably, concerning the practical manner to know how to execute an action, two main options are available in the recent debate: 3
Option a. Some philosophers state that knowing how to perform F in a practical manner involves nothing but the possession of a suitable propositional content about the way to execute that action, accessed in a practical mode of presentation (cf. Stanley & Williamson, 2001). According to this view, knowing how to perform an action in a practical manner involves representing the way to execute that action in a distinctively practical sense (Pavese, 2015, 2017).
Option b. Others maintain that knowing how to perform F in a practical manner requires a content about the way to execute F which has a non-propositional format (e.g., Ferretti, 2020; Levy, 2015; Pacherie, 2008). According to this view, knowing how to perform a certain action in a practical manner entails the presence of a motor representation which guides the execution of the intended action (Fridland, 2013, 2016, 2021).
Options a and b both entail that in order for Ariel to know how to perform action F in a practical manner, something over the mere knowledge in the epistemic sense must be possessed by Ariel (see also Ryle, 1949, pp. 38–39). More precisely, if option a is true, it makes sense to say that Ariel knows how to F in a practical manner only if Ariel has the (true and justified) propositional representation that w is a way to execute F accessed in a practical mode of presentation. Differently, if option b is true, it makes sense to say that Ariel knows how to F in a practical manner only if Ariel has the motor, nonpropositional representation of the way w to perform F.
When investigating the practical manner by means of which we know how to execute an action, the philosophical debate has long been divided regarding the adequacy of options a or b. 4 We overcome this specific dichotomic debate by showing that knowing how to perform F in a practical manner involves both a propositional and a nonpropositional, motor representation of the way w to execute F. These two states can operate together in a representational dance mixing different representational formats. How?
In brief, our proposal is that the propositional knowledge that w is a way to F, accessed in a practical mode of presentation, is constituted by the executable action concept that activates the motor representation of the execution of F in the way w. Accordingly, the propositional knowledge that w is the way to F guides the execution of F, if the action concept that constitutes the propositional content of such knowledge activates the relevant motor representation that allows the execution of F in the way w.
In this respect, in order to answer our question about what mental states underlie our knowing how to execute an action in a practical manner, we rely on the following assumptions:
Fregean construal: the practical mode by means of which the subject has access to the propositional knowledge that w is a way to F depends on the representation of the action concept of F that constitutes that content. 5
Executability of action concepts: the representation of the action concept of F pertains to the class of executable action concepts if it activates the appropriate motor representation of F, so as to allow the subject to form the appropriate intention to F in a certain way. 6
By combining assumptions (1) and (2), we conclude that:
3. The propositional knowledge that w is a way to F, accessed in a practical mode, allows the subject to intend to perform F in a certain way w, if the action concept of F that constitutes this knowledge state is the executable action concept that activates the motor representation of F in the way w.
In this framework, we have a unified criterion to solve our problem about Ariel’s knowledge: Ariel’s propositional knowledge that w is the way to execute F accessed in a practical mode entails that Ariel knows how to execute F in the way w, according to a practical manner, if such a propositional knowledge is constituted by the relevant executable action concept, that is, by Ariel’s motor representation of the execution of F in the way w.
We now need to defend this idea with a look at the literature. Our argument proceeds in two steps. First, we focus on the notion of practical mode of presentation to account for knowing how to F in a practical manner. In this respect, we will discuss the seminal works by Stanley (2011) and Stanley and Krakauer (2013), arguing why their proposals are not able to account for the flexibility that characterizes the execution and the control of an action. This makes these proposals unsuitable to explain the way our knowledge guides our actions. Then, we focus on the recent proposals by Pavese (2015, 2017, 2019), according to which a practical mode of presentation should be intended as the representation of the motor procedure that is prescribed by means of the intention to execute an action. More precisely, in this part we argue that, since Pavese (2019) assumes that motor procedures have a prescriptive meaning (pp. 805–806), she opens the door to the role of a nonpropositional component in practical knowledge, inasmuch as prescriptions do not have truth-values. 7 However, if accessing the propositional content of our knowledge in a practical mode implies reference to a nonpropositional component, then it is necessary to explain how such nonpropositional component interfaces with the propositional intentions that underlie such an action. Our goal is precisely to take a step forward and fill this gap, thus successfully extending the account by Pavese by overcoming the problems remaining unsolved in her framework.
To do this, in the second part of this article, we introduce the Same Format Thesis (SFT), as developed by Ferretti and Zipoli Caiani (2021), and suggest that it can help us in understanding how propositional knowing how to execute an action in a practical manner hooks up with the nonpropositional representation of the procedure that allows a subject to form the appropriate intention to execute this action. Notably, we will provide evidence that executable action concepts are represented by the agent through the same (neural) realizers that the agent uses to prescribe the execution of a motor plan for action. Thus, we drawing on the SFT, we provide arguments to maintain that the representation of executable action concepts and motor representations share a common representational format.
The picture we obtain from our account is the following: an agent knowing that w is a way to F, knows how to execute F in a practical manner, only with the proviso that they have the executable action concept allowing them to form the appropriate intention to F in the way w. Now, since an executable action concept is an essential component of a practical proposition (Pacherie, 2008, 2011), and since having an executable action concept amounts to possessing the appropriate motor representation for the associated action, we maintain that knowing the proposition that w is a way to F, in a practical manner, rather than in an epistemic manner, involves the role of a nonpropositional motor representation as its constituent. Importantly, our proposal leads us to take a step forward in understanding the relationship between the propositional content of our knowledge and the nonpropositional content of our motor representations.
Practical modes and the selection–prescription issue
Stanley and Williamson (2001) have argued that knowing how to execute an action is best construed as entirely propositional. Knowledge how is not a peculiar type of knowledge, but only a branch of propositional knowledge. One knows how to F in a way w, if and only if there is some relevant way w such that the agent stands in the knowledge relation to the proposition that “w is a way to F,” and w is presented according to the relevant practical mode (Stanley & Williamson, 2001, p. 430). But why should there be something like a practical mode of presentation?
Several scholars have expressed doubts about invoking the notion of practical mode, arguing that this notion is too vaguely introduced (Levy, 2015; Noë, 2005), or contending that it is not a viable route to explain how knowledge guides action (Dickie, 2012; Ferretti, 2020; Fridland, 2013; Glick, 2013).
Here we focus on the problem of understanding whether propositionally knowing that w is a way to F, accessed according to a practical mode, suffices to knowing how to execute F in the way w in a practical manner. Notably, in order for the propositional knowledge that w is a way to F to guide the execution of F in a practical manner, it must be able to generate the suitable intention to execute F in the way w. So the question is: how does our propositional knowledge that w is a way to F generate the intention to execute F in the way w?
Note that one can intend to execute F through different movements, involving a plethora of action outcomes and related intentions. Indeed, one can intend to hold a mug, in order to drink, in several ways, that is, by means of different types of grips, each of which being performed through different motor outcomes (Ferretti 2016, 2021; Jacob & Jeannerod, 2003; Jeannerod, 2006). For example, one can intend to hold a mug either through a precision grip, or with a power grip, and either grasping the handle, or the border of the mug, either with the left or the right hand. If so, merely knowing that performing a precision grip is a way to hold a mug is not enough in order to select and prescribe the suitable action outcome that allows one to hold the mug in a certain way.
Things being so, if the practical mode of presentation is an ingredient needed to form the intention to execute the related action, a twofold issue arises. First, it should be clarified how knowing a proposition according to a practical mode selects a specific action outcome among others. Second, it should be clarified how accessing a propositional knowledge according to a practical mode contributes to the prescription of precisely that motor chain of movements that constitutes the selected action outcome. This is what we call here the selection–prescription issue.
In the following subsections, we focus on two proposals concerning the notion of practical mode of presentation and how it relates to the selection–prescription issue. First, we discuss the approach developed by Stanley (2011) and Stanley and Krakauer (2013). Then we consider the neo-Fregean view introduced by Pavese (2015, 2019, 2021). We show that, although the selection–prescription issue has been addressed in the literature, a step further can be still advanced in the solution of this problem.
Automatic mechanisms and the problem of flexibility
Stanley (2011) famously holds that practical knowledge interlocks with a motor outcome by means of an automatic mechanism, and that the translation of propositional knowledge into action does not itself require further knowledge. Stanley defends the idea that, in order to put propositional content into action, a dedicated cognitive module performs the function of selecting and triggering the appropriate motor outcome. Cognitive modules are usually intended as specialized tools characterized by a fixed architecture, so that the operation of a module cannot be altered by other concomitant processes in the system. According to Stanley, cognitive modules help to avoid the infinite regress as famously addressed by Ryle (1949), since in this way no independent grasp of meaning is required in order to mediate between epistemic states and actions (Stanley, 2011, p. 16).
In another paper, Stanley and Krakauer (2013) argued that selecting and triggering the suitable representation of the motor outcome for action execution does not rely on propositional knowledge only, but also on a component that is not knowledge-based, namely, the procedural aspect (p. 8). This procedural component requires motor acuity, an automatic and fixed disposition to react in a certain way developed through practice.
For both accounts, knowing how to do something in a practical mode involves the role of an informationally insensitive component, which functions as a mediator between the propositional content and the representation of the relevant motor outcome. This component carries out its function without being semantically or inferentially related to other independent information available to the agent, as otherwise it could not be either a cognitive module or an automatic behavioral disposition as Stanley (2011) and Stanley and Krakauer (2013) respectively recommend.
As noted by Levy (2015) and Fridland (2013, 2016), the idea of framing the selection–prescription issue in a way that is not sensitive to information suffers from a flaw. Indeed, for an automatic mechanism to fill the function, it must be responsive to the relevant attributes that influence the online control of an action. To play such a role in action control, practical knowledge must trigger the representation of the related motor outcome intelligently, namely, in a manner that is sensitive to contextual information. This happens when one knows that w is a way to F, and because of this knowledge, one can execute instances of F in a way w. For example, a golf player may know that holding a club, say a driver, with a specific grip, is a way for them to perform a successful shot, and despite them using this type of grip several times during a match, they may give rise to motor outcomes characterized by differences in the degree of force, precision, and angle. Such a variety of action outcomes commonly depends on the available information, such as individual considerations concerning the characteristics of the environment, the distance of the green, the properties of the ground and the direction of wind, as well as on individual aspects, such as the strategy adopted by the player, their level of tiredness, and many others (Fridland, 2013, 2016).
This shows that selecting and triggering motor outcomes cannot be framed as an informationally insensitive process. Indeed, were a module or a mere behavioral disposition responsible for selecting the motor output of an action, it would not be possible to adapt the execution of this action to the different contexts in which it can be performed. This is because cognitive modules and automatic behavioral dispositions have a fixed structure, whose function cannot be affected by any change in the flow of internal and external information (Levy, 2015).
This consideration has consequences also for the practical mode of accessing propositional knowledge. If knowing a proposition according to a practical mode is the ingredient by means of which to account for knowing how to do something in a practical manner, it follows that the practical mode to access propositional knowledge cannot be framed in terms of informationally insensitive mechanisms. Rather, for a propositional content to properly interface with the appropriate action intention, the practical mode according to which it is accessed needs to be conceived in a representational way, that is, a way sensitive to the information concerning the internal and external environment. This view has been endorsed by Pavese (2019), who has framed the notion of practical modes in terms of prescriptive representations. We will address this in the next section.
Practical modes and the compositional framework
Pavese (2015, 2017, 2019) worked out a notion of practical mode of presentation as a practical sense in a Fregean construal, that is, as a constitutive element of a propositional structure. Notably, when conceived from a Fregean perspective, a practical sense is a concept construed as to be the mode of presentation of a particular way to execute an action (Pavese, 2015, p. 2). Pavese (2015) maintains that knowing how to do something by means of a practical mode amounts to know a given propositional content, which one can be acquainted with according to a practical sense. Now, since on this view practical senses are conceptual elements, à la Frege, then they are representational vehicles of some sort. Accordingly, it follows that propositional contents interface with action by means of a representational vehicle.
Notably, practical senses are ways of representing motor commands to be executed. If one represents how to execute an action according to a practical mode of presentation, one represents the instructions that they have the ability to execute. Representing how to F in the way w according to a practical mode involves the ability to perform F in the way w (Pavese, 2017, 2019).
Interestingly, Pavese provides details of what she calls a two-dimensional model of motor commands semantics. Motor commands are vehicles of information endowed with a twofold representational function: on the one hand, they denote an action to be executed, on the other, they prescribe the execution of that action. Thus, while a motor command denotes an action having an outcome as a referent, it prescribes the execution of this outcome expressing a sense according to which it must be performed in a certain way. As Pavese (2019) writes: “practical representations in the sense discussed here are prescriptive: they represent a task as to be performed in a certain way” (p. 801).
However, since according to this model motor commands are mental states endowed with a prescriptive function, they are intensional (opaque) entities. Prescriptive representations are intensional because the same action can be denoted by a set of motor commands characterized by different senses, that is, by different ways to prescribe a manner in which we can perform that action. For example, we already said that one may refer to an action, say, holding a mug to drink, by means of different motor commands, each of which prescribes the execution of this action whereby different motor outcomes are characterized by the use of different effectors or types of grips. This is tantamount to considering motor commands as perspectival representations related to different modes of presentation of the same referent (Pavese, 2019, p. 797).
Moreover, according to Pavese (2019), prescriptive representations have a nonpropositional content. 8 A prescriptive representation, indeed, is more like an imperative (p. 793) than a description. Therefore, it does not contain the content conditions of truth. As a consequence, the representation of a motor command does not provide information in a propositional structure, for propositions are defined essentially as objects characterized by truth conditions.
To sum up, knowing how to execute an action according to a practical mode of presentation is analyzed as a composition of two elements: a declarative element, corresponding to the knowledge of a propositional content, and a procedural element, corresponding to the prescriptive representation that constitutes the propositional content (Pavese, 2019, p. 799). Now, since prescriptive representations have intensional and nonpropositional structures, this compositional framework involves that knowing how to do something in a practical mode requires the possession of a propositional content, which is constituted by an intensional, nonpropositional representation. This point will be of key importance for our claim.
Before concluding this section, we analyze carefully the selection–prescription issue, that is, the problem of understanding how propositionally knowing that w is a way to F according to a practical mode can appropriately trigger the suitable intention to execute F in the way w.
First, the compositional framework involves an interaction between propositional knowledge and action intentions in terms of representational contents. This allows avoiding the main flaw affecting the proposals by Stanley (2011) and Stanley and Krakauer (2013). If so, propositional knowledge interfaces with action performance by means of informationally insensitive mechanisms, which are not able to account for the flexibility that characterizes the ways of intending the execution of an action. In other words, according to the compositional framework, a proposition about how to execute an action may refer to a variety of action outcomes depending on the procedural element associated with it. Thanks to the intensionality of practical senses, knowing the proposition that w is a way to F in a practical mode allows one to form the intention to perform F in the way w by means of different manners of execution. This is exactly the flexibility that the selection–prescription issue requires in order to be solved.
Based on these considerations, there is a reason to prefer the compositional framework over modularist and dispositionalist views. The former can account for how propositional knowledge interfaces with action intentions in a flexible way, whereas the latter are not able to do the same. Now, knowing, for an agent, how to do something in a practical mode is a matter of having the right procedural element that prescribes the manner by means of which to execute F. But this notion further begs the question: what exactly is such a procedural element?
According to Pavese (2019), a procedural element is a nonpropositional representation, which can be conceived as a motor representation (Pavese, 2019, pp. 797–800). Knowing a proposition about how to execute an action in a certain way, say, how to grab a bottle using the related motor representation of that action, “is just one way of knowing a proposition under a practical mode of presentation” (Pavese, 2019, p. 805). This claim, however, can raise some concern. While Pavese (2018, 2019) analyzed the relation between propositional contents and practical senses, there is still the need to explain how motor representations are integrated with propositional knowledge.
The next section provides this explanation, suitable for solving the so-called interface problem.
The motor representation of (executable) action concepts
Explaining how practical senses properly relate to propositional knowledge (through motor representations) amounts to explaining the way propositional and motor states can interlock. This, in turn, entails solving what has been recently called the interface problem which has attracted the attention of many philosophers. The problem asks about how intentions, having a propositional formatted content, and motor representations, which have a differently motorically-formatted content, interlock in the selection of proper action performance (Burnston, 2017; Butterfill & Sinigaglia, 2014; Fridland, 2021; Mylopoulos & Pacherie, 2017; Shepherd, 2019). Solving the interface problem provides then, ipso facto, a solution to the more general question concerning the integration of motor representations and propositional knowledge.
We focus, in this respect, on a recent solution to the interface problem, the Same Format Thesis (SFT), according to which propositionally structured intentions and motor representations interlock through the physical realizers subserving the representation of executable action concepts (Ferretti & Zipoli Caiani, 2018; cf. Ferretti & Zipoli Caiani, 2021). Consider that the representation of executable action concepts is functionally related to plans for action, which cause, in an agent, the execution of the corresponding series of goal-related movements (Israel et al., 1993; Pacherie, 2011; Pavese, 2021). This means that for an agent having the representation of an action F as an executable concept involves the possession of the suitable mental state that prescribes the production of F, namely, the related motor representation of the action F.
The SFT provides a framework to understand how a motor representation can be related to the propositional content of an intention, by describing how the processing of motor representations and the processing of the executable action concepts that constitute the intentions share the same physical realizers (Ferretti & Zipoli Caiani, 2018). Then, an agent can propositionally intend to perform the action F precisely because they possess the executable action concept of F which itself is realized in the system as the motor representation of F.
Although the SFT has been originally introduced to account for the interlock between propositional intentions and motor representations, as to solve the interface problem, it can be used here to understand the more general question concerning how motor representations can be involved in propositional knowledge in the light of its relation to practical senses, so as to offer a step forward for the literature on practical knowledge. In brief, in what follows, we argue that when one knows that w is a way to F in a practical manner, then the agent is representing F by means of an appropriate executable action concept, this amounting to possessing the motor representation of the execution of F in the way w. 9
On concepts and executable actions
The main argument supporting the SFT is based on the assumption that the executable concepts of actions and the related motor representations share the same realizers in the cognitive-motoric system. This argument relies on three types of well-established evidence (here we provide only a taste of the empirical findings supporting SFT; for an accurate review, see Ferretti & Zipoli Caiani, 2018, 2021).
First of all, there is behavioral evidence according to which the processing of executable action concepts modulates the execution of the related action (Andres et al., 2015; Boulenger et al., 2006; Gentilucci et al., 2000; Glover et al., 2004; Klepp et al., 2017), whereas , vice versa, the execution of an action affects the processing of the related executable concept (Lindemann et al., 2006; Rueschemeyer et al., 2010; van Dam et al., 2012; van Elk et al., 2008). Importantly, behavioral results reveal that the acquisition of novel skills improves the ability to recognize and classify different types of actions, suggesting that the representation of executable action concepts maps onto the representation of motor instructions (Casile & Giese, 2006; Glenberg et al., 2008; Iverson, 2010; Locatelli et al., 2012).
The second piece of evidence supporting SFT is about the neurobiological processing of the executable action concepts. Interestingly, a number of results have established that the cortical motor system is somatotopically activated during the understanding of action verbs. For example, Ghio et al. (2018) showed that the processing of action-related verbs induces congruent modifications in those regions of the brain that code for the motor actions they are referred by the verbs. Moreover, many works have shown that the processing of different action concepts relies on the motor areas involved in the execution of the actions they refer to (see also Boulenger et al., 2009; Coello & Fischer, 2015; Desai et al., 2010; Fargier et al., 2012; Hauk et al., 2004; Kemmerer et al., 2008). Crucially, it has also been shown that the activation of the motor system in the processing of action concepts is a translinguistic and transcultural phenomenon (Pulvermüller, 2013; Wu et al., 2013; see also Beilock et al., 2008; Lyons et al., 2010; Tomasino et al., 2013).
Lastly, the use of functional techniques has shown that the representation of action concepts is part of the representational activity of the motor system, since the specific impairment of this cortical region is associated with difficulties in using and understanding action-related sentences (Bidet-Ildei et al., 2017; Desai et al., 2015; Fernandino et al., 2013). Indeed, subjects with lesions in their motor system present higher level cognitive deficits involving a weakening in the processing of action concepts (Bak & Chandran, 2012; Desai et al., 2015; Ibáñez et al., 2013; Kemmerer et al., 2012). This crucially suggests that the activity of the motor system is not a mere consequence of action concepts processing but, on the contrary, that it also has a deep functional role in the representation and computation of the executable action concepts that constitute propositional structures. Notably, considering the recruitment of motor representations only a consequence of the processing of amodal action concepts would not explain also the impairment of the agent’s information processing about motor acts as compromising the ability to process action concepts. Crucially, the representations of action we are talking about here, at the level of both the processing of action concepts and the processing of motor representations, are related to very specific forms of action (i.e., “grasp the mug with a precision grip”).
Summing up, evidence shows that the representation of executable action concepts is grounded within the sensorimotor system, and that this depends on the mode of execution that is proper for the type of action they relate to. Since evidence shows that the representation of executable action concepts and the related motor representations, at least in the considered cases, share the same physical realizer, it is reasonable to suppose that, in these cases, they also share the same representational format, which is non-propositional, this being in tune with the literature (Butterfill & Sinigaglia, 2014; Jeannerod, 2006; Nanay, 2013; Pacherie, 2018). Let us go more slowly on this.
The same format thesis
In the previous section, we provided evidence that the executable concepts of actions and the related motor representations share the same realizers in the cognitive-motoric system. Now, we provide reasons to endorse the claim that, since executable action concepts and motor representations share the same realizers, they also share the same representational (motor) format. This is the core of the SFT.
Before addressing the SFT, however, it should be remembered that, according to the literature, the format of a representation is the peculiar way through which a vehicle delivers a content, that is, the structure of the representation (Butterfill & Sinigaglia, 2014; Levy, 2015; Nanay, 2013). A content, indeed, can be delivered in different ways: with propositional structures, pictorial structures, schematic structures. Take the case of a route, say, the path leading to your preferred coffee shop. Notably, not only can there be different routes that take you there, but each of these routes can be represented by means of different structures: as a series of propositional descriptions, as a depicted sign on a map, or as a schematic sequence of symbols.
Representational systems having different formats can be associated with “characteristic performance profiles” (Butterfill & Sinigaglia, 2014, p. 126). We can identify a representational format on the basis of the function of the process it appears in. For example, propositionally formatted representations feature in discursive reasoning, pictorial representations foster similarity analysis concerning images, while schematic structures can be found in instructions. Now, since the function of the process generated by a representational system depends on its format, representational systems having different formats perform different functions. This means that (a) representations having different formats have dissociable functions, while (b) representations sharing the same function may have a common representational format.
This should be compared with the evidence that executable action concepts and motor representations share the same realizers, and particularly with the fact that subjects with lesions in their motor system display higher level cognitive deficits leading to an impairment in the processing of executable action concepts: that is, same neural impairment, same functional impairment (Bak & Chandran, 2012; Desai et al., 2015; Ibáñez et al., 2013; Kemmerer & Gonzalez-Castillo, 2010). Evidence leads to stating that the representation of executable action concepts and motor representations share a common realizer. Hence, they share a common causal function, and therefore a common representational format (according to our point b, above). After all, representational functions performed by the same neural realizers are related to similar representational formats, for the reasons exposed above. The representational format is crucial for the way in which the function is shaped and, accordingly, the content represented.
Importantly, the hypothesis that executable action concepts are encoded in a motor format explains the available evidence better than the hypothesis according to which executable action concepts and motor representations rely on separate mental substrates and formats (Ferretti & Zipoli Caiani, 2018, 2021). Again, if the executable action concepts and the related motor representations were encoded by different representational formats, there would be no reason to expect that an impairment of the agent’s motor system would affect the functional processing of action categories. This is sufficient, we think, to endorse the SFT.
The relevance of the same format thesis for knowing how to perform an action
Coming back to our main issue, the SFT is crucial in explaining how the executable action concept related to performing a given action F in the way w can be linked to the propositional knowledge that w is a way to execute the action F. Indeed, it provides an account for the interlock between motor representations, through executable action concepts, which are endowed with a nonpropositional format, and propositional knowledge, which is endowed with a propositional format.
More precisely, it explains the way in which the epistemic component that characterizes our knowing how to perform an action relates to the practical component that allows us to execute that action. It does so by accounting for the interlock between propositional structures and nonpropositional motor-representations when showing that the propositional representation of our practical knowledge is constituted by the relevant executable action concept, which is realized by the related nonpropositional motor-representation. But this crucially allows us to understand why propositional knowledge accessed according to a practical mode of presentation is knowledge in the relevant, practical manner. 10
At this point, it is crucial to realize that framing practical knowledge within the SFT framework has three crucial advantages for a compositionalist approach such as that of Pavese (2019; and ours). First, and most important, it provides a solution to the selection–prescription issue, by explaining why knowing that w is a way to F according to a practical mode of presentation leads one to form a suitable intention of the action, which is constituted by the motor representation responsible for the execution of the action F in the way w. This is possible by assuming that a propositional content, accessed in a practical mode, has an executable action concept as its constituent, and that this constituent has the same realizer of the motor representation that forms the related intention to act. Now, since according to the SFT, the executable action concept figuring in an intention has the same realizer of a motor representation, it follows that knowing how to perform an action in a practical manner requires the possession of the motor representation that is needed to form the intention to execute that action. Therefore, a consequence of the SFT is that propositionally knowing how to perform an action in a practical manner involves the ability to form the intention that is constituted by the suitable motor representation responsible for the execution of such an action. 11
That is, if I know that w is a way to F, and if I intend to F in a way w, the representation of the concept of the action F, which is part of the propositional structure of both my intention and my propositional knowledge, interlocks with the appropriate motor representation. Then, knowing F in a practical mode amounts to having the representation of an executable action concept that shares the same realizer of the corresponding motor representation of the way to execute F. Thus, executable action concepts are the mental items allowing us to have a given w to F known in a practical mode. If so, their involvement in a propositional structure, along with them being built in a motor format, allows us to have propositional knowledge in a practical mode of presentation. This leads to recruiting the appropriate motor representations, since the executable action concept that constitutes the propositional knowledge and the related motor representation share the same realizer and format. Then, the mechanism described above clearly suggests a solution to the selection–prescription issue.
The second advantage of the SFT is that it saves the intensionality of the action concepts required in practical knowledge. According to Pavese (2019), since there are different ways to prescribe a way to perform an action, the prescriptive component that constitutes the practical mode in which one knows a practical content is an intensional (opaque) representation. In a Fregean framework, developed by Pavese, indeed, there are many practical senses by means of which an action can be denoted, this depending on the action outcome that an agent wants to satisfy by executing as given action. But this falls short of an explanation of how specific prescriptions, related to opaque intensions, can recruit specific actions in a given motor context.
Consistently with this assumption, the SFT allows us to understand how propositionally knowing that w is a way to F, according to a practical mode, can be coherently related to a specific motor action, thanks to the interlock between a given executable action concept, contained in a very specific intention, and a given motor representation. Since the representation of the executable action concept of F and the motor representation of F share the same realizer, and since different motor representations of F prescribe different ways to execute F (Pacherie, 2018), the action concept of F may have different practical senses as a function of the motor representations that realize that concept. It is then because the executable action concept that constitutes a propositional structure is subserved by the same representational vehicle and, ipso facto, realized with the same format of a motor representation (see Ferretti & Zipoli Caiani, 2018, 2021), that knowing according to a practical mode of presentation can have different practical senses, but, still, only one action is selected. Therefore, knowing how to F in a practical mode can, in principle, trigger different ways of intending the execution of F, this depending on the motor representation that instantiates the executable action concept of F in the system. But, in a peculiar motor context, the specific interlock between the representation of a given executable action concept and a specific motor representation allows us to select a suitable action.
Finally, the third advantage is that the SFT accounts for the flexibility that characterizes the interlock between knowing how to perform an action and the related action outcomes. As noted by Levy (2015) and Fridland (2013, 2016), to play its role in action control, practical knowledge must trigger the representation of the relevant motor outcome intelligently, namely, in a manner that is sensitive to contextual information. The SFT accounts for the intelligent way according to which propositionally knowing how to perform an action can give rise to (exactly one intention of an action among) the variety of modes of intending the execution of action, precisely as a function of contextual information (Ferretti & Zipoli Caiani, 2021). This is because motor representations that are at the basis of executable action concepts are not segregated from other processing of such intelligent information, but rather are deeply integrated in higher forms of cognition, such as those concerning perceptual and semantic processing (Ferretti, 2016; Ferretti & Zipoli Caiani, 2021; Jacob & Jeannerod, 2003; Jeannerod, 2006).
Consider also that part of the processing subserving motor representations is subserved by those areas involved in manipulating information for categorization and semantic purposes. This aspect of motor representations clearly explains their relation to higher forms of processing, which can be related to semantic, categorical representations of the information that must be computed for suitable action performance (for a review, see Briscoe, 2009; Ferretti, 2016; Jacob & Jeannerod, 2003; Nanay, 2013; Zipoli Caiani & Ferretti, 2017). 12
Summing up, contrary to the hypothesis that practical knowledge interlocks with action through an automatic and informationally insensitive mechanism (Stanley, 2011; Stanley & Krakauer, 2013), the SFT accounts for the flexibility of action control and execution in an intelligent manner, in line with the literature (Fridland, 2013, 2016). Indeed, since the executable action concept that forms a propositional content known according to a practical mode of presentation is instantiated in the system as a motor representation, and since motor representations are sensible to semantic information, it is now clear how and why practical knowledge can interlock with action in a flexible way, this further extending the description of practical modes (Pavese, 2015, 2017).
Conclusion
This article provides a framework capable of accounting for the way in which our knowing how to execute an action can effectively guide the actual execution of our actions. This is done starting from two main premises. First, under a Fregean construal, the propositional knowledge that w is a way to F accessed according to a practical mode is constituted by the representation of the action concept of F which one can be acquainted with according to a practical sense, that is, according to the way of representing the relevant motor commands to be executed. Second, in order for the action concept of F to represent a motor command to be executed, it must pertain to the class of executable action concepts which activate the appropriate motor representation of F. By combining them, we have argued that the propositional knowledge that w is a way to F, accessed in a practical mode, leads to intend to perform F in the way w if the action concept of F that constitutes this intention is the executable action concept that activates the motor representation of F in the way w. To support this claim, we have drawn on the SFT, which suggests how the executable action concepts contained in the propositional structures that form our knowing-how interlock with the motor representations capable of satisfying an intentional action in a specific motor context. Importantly, our proposal allows us to take a step forward in answering our very initial question concerning the relationship between the contents of our knowledge states and our ability to execute motor actions.
Footnotes
Authors’ Note
This article was initially conceived, deeply discussed and finally written by means of a shared work of both authors, so much so that it is impossible to attribute a particular merit to one or the other author, given the deep interplay at the basis of such a shared effort. However, having to recognize responsibilities in case of administrative reasons, we can say that Silvano Zipoli Caiani wrote the first part, which was meticolously reviewed, amended and finally approved by Gabriele Ferretti, while Gabriele Ferretti is responsible for the second part, which was meticulously reviewed, amended and finally approved by Silvano Zipoli Caiani.
Authors are listed in alphabetical order
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: Gabriele Ferretti would like to acknowledge support from a Humboldt Fellowship, hosted by Professor Albert Newen at the Institute for Philosophy II, Ruhr- University Bochum, Germany.
