Deweyan conceptual engineering: reconstruction, concepts, and philosophical inquiry

ABSTRACT

Reconstruction is a central notion in Dewey’s account of inquiry and in his metaphilosophical commitments. In his work, Dewey made a call for reconstruction of philosophy, in the reconstruction of central notions of the discipline, like knowledge, logic, truth, the good, reason, and experience. Inquiry itself is reconstructive, according to Dewey, involving the transformation of an indeterminate situation into one which is determinate and understood. Dewey’s philosophical views should therefore be of interest to those taking part in the recent turn towards revisionary philosophical methodologies, like conceptual engineering, explication, and amelioration. In light of the recent developments in revisionary methodologies, I aim to explore Dewey’s conception of concepts in relation to the conceptual engineering literature, suggesting that it provides a useful conception of concepts for conceptual engineering, and to suggest a Deweyan model of philosophical inquiry where concepts are the objects of inquiry. In both cases, I hope to show a productive interaction between Dewey’s ideas and contemporary discussions of conceptual engineering.

KEYWORDS:

• John Dewey
• pragmatism
• conceptual engineering
• inquiry
• reconstruction

1. Introduction

Reconstruction is a central notion in Dewey’s account of inquiry and in his metaphilosophical commitments. In his work, Dewey made a call for reconstruction of philosophy, in the reconstruction of central notions of the discipline, like knowledge, logic, truth, the good, reason, and experience. Inquiry itself is reconstructive, according to Dewey, involving the transformation of an indeterminate situation into one which is determinate and understood. Dewey’s philosophical views should therefore be of interest to those taking part in the recent turn towards revisionary philosophical methodologies, like conceptual engineering, explication, and amelioration. In light of the recent developments in revisionary methodologies, I aim to explore Dewey’s conception of concepts in relation to the conceptual engineering literature, suggesting that it provides a useful conception of concepts for conceptual engineering, and to suggest a Deweyan model of philosophical inquiry where concepts are the objects of inquiry.

Dewey thought of concepts as rules for actions or operations to be performed. Concepts are not final or static, but integral tools in the dynamic activity of inquiry. In explicating this idea, Dewey thought of the meaning of concepts in terms of what he calls the means-consequence relation. There are benefits to draw from this view of concepts for thinking about conceptual engineering (CE). In this paper, I will (1) demonstrate how this view of concepts provides the grounds for actionable CE when combined with (2) Deweyan insights about the structure of inquiry. Points (1) and (2) are related since (2) helps demonstrate the strength of (1), while (1) provides the grounds for a re-conception of philosophical inquiry through the Deweyan conception of concepts (DCC). Both points provide us with insights from Dewey’s pragmatism that can help us do CE better, while also providing a new contemporary lens through which to view Dewey’s work.

In Section 2, I introduce the recent discussion of concepts of concepts for conceptual engineering, focusing on Manuel Gustavo Isaac’s (2021) comparison of the philosophical and psychological concepts of concept. I agree with Isaac’s assessment of the philosophical concept of concept and the superiority of the psychological notion. However, after formulating the DCC in Section 3, I argue in Section 4 that the DCC can take the benefits of the psychological concept further, while also avoiding the problem of psychologising concepts. First, and most importantly, the DCC makes CE actionable since concepts bear out their significance in the actions we perform (‘action is at the heart of ideas’ LW4, 134). Second, it has a broad scope and impact, in that it captures the importance of concepts in the cognitive – and practical – aspects of our lives. Third, as with operationalist thinking about concepts, it provides clear definitions and rules for use of the concept. The comparison with Isaac’s psychological concept is apt, since Dewey himself is clear that his ‘conception of conceptions’ is itself subject to the experimental criterion on knowledge and inquiry; that is, it must be fruitful and prove its worth in practice (LW4, 118n2), like Isaac (2021) views the discussion of concepts itself as a CE project. In Section 5, I discuss the relationship between the DCC and Dewey’s account of the structure of inquiry, providing the foundations for an active account of CE.

2. Concepts of concept for conceptual engineering

With the amount of attention garnered by CE in recent years, there has also been a proliferation of concepts of concepts. Isaac (2021) discusses two distinct notions of ‘concept’ for the purposes of CE: the philosophical concept of concept and the psychological concept of concept. The philosophical concept – which Isaac notes is typically adopted in CE – makes sense of concepts as the components of the contents of our thoughts, while serving ‘a referential function’ (Isaac 2021). Concepts, on this view, have some analytical and a priori semantic structure, ‘along with […] application criteria to the objects that fall within its extension’ (Isaac 2021). The primary function of concepts on this view is to pick out or denote objects in its extension. As Cappelen (2018, 183) puts it, ‘the only universal, i.e. stable, function of a concept “C” is to denote Cs’. The benefit of the philosophical notion of concepts can be seen in the way it accounts for stability (and objectivity) of meaning and same-saying: concepts, utterances, and thoughts are the same if they are about the same things, if they have the same extension or denote the same objects.

The psychological concept, on the other hand, identifies concepts as

structured bodies of information about some category of referents […], which are retrieved and activated, or constructed, to play a causal/explanatory role in the cognitive processes that underlie most of our higher cognitive competences in relation to these referents. (Isaac 2021, §4)

Concepts, on this view, are individuated on the basis of the functional roles they play in various cognitive tasks and are further evaluated based on their efficiency in those tasks. Contrasted with the clear stability of the philosophical concept, psychological concepts look to be more malleable, since a concept is identified and can be revised based on the kind of role it plays. While the philosophical concept might be able to account for the stability required for same-saying, it does also make the idea of CE difficult to grasp, since any change in concept will require a change in extension. The philosophical notion, as Isaac also argues, is too static – it makes conceptual change and CE impossible, or at least very difficult (cf. Cappelen 2018, chs. 6 and 7). The psychological concept, on the other hand, makes sense of concepts in functional terms, and makes CE more than a mere possibility: it makes it actionable and useful, since changing a concept will be a matter of changing minds.¹ As I will argue below, this also causes some issues for the psychological concept, in being too psychologistic.

Isaac (2021) makes a convincing case for the superiority of the psychological concept over the philosophical concept by comparing them along different dimensions. The three dimensions are adapted from Carnap’s (1950) assessment of explicatory projects in terms of fruitfulness, exactness, and simplicity. For Isaac (2021), fruitfulness is explicated along the lines of the scope and impact concepts have on our cognitive lives. The scope of a concept of concepts will be evaluated in terms of how well it can capture the breadth of concepts, ranging from theoretical to practical, while the impact of a concept of concepts is determined by the way it allows CE to ‘affect people’s minds and cognitive lives’ (Isaac 2021, §3). Exactness is explicated in terms of unambiguity, precision, and consistency. The latter is determined by the lack of inconsistent constituent principles and absence of paradoxes derived from those principles, while unambiguity and precision are a matter of descriptive or explanatory accuracy and adequacy respectively. In the case of precision, descriptive accuracy is measured by theoretical convergence, while in the case of unambiguity adequacy is determined by positive theoretical results. Finally, simplicity is a matter of definition and rules of use for the concept, where the former is a matter of providing operationalisable concepts that the latter can make use of for the purpose engineering and the implementation of the proposed engineered concept.

The major problems facing the philosophical concept can be summarised through the way in which it fails to provide operationalisable concepts for the work that it aims to do (failing simplicity, as explicated by Isaac); it makes CE difficult or impossible, both in the sense of providing too rigid a notion of concept and in terms of having little to do with the cognitive lives of humans (since concepts’ primary function is to pick out their extensions); it has little to show for it empirically, failing ‘to identify the basic constitutive features of concepts’ (Isaac 2021, §5.2); and so provides little to no guidance on operationalising and identifying concepts in experiments and practice. The psychological concept, on the other hand, rectifies these problems by focusing on identifying concepts with the cognitive and functional roles they play in people’s lives. The psychological concept is grounded in successful, replicable empirical and experimental psychological research, in contrast with the philosophical concept. The psychological concept thus has a more promising track-record than the philosophical concept, given the empirical successes it has experienced (cf. Isaac 2021, §5.2).

While the psychological concept focuses on changing minds, the DCC I formulate in the next section focuses on changing actions: implementations of concepts is a matter of changing people’s actions, and not just their minds. Indeed, Dewey would have been worried about psychologising concepts because it seems to make concepts private. Making concepts a matter of something psychological makes them private and ineffable, leaving it mysterious how one of the most essential functions of concepts – communication and coordination of action – is achieved (Dewey LW12, ch. 3; Gronda’s 2020, 67–69). However, Dewey would be equally worried about the hypostatisation of meanings that occurs on the philosophical conception of concepts (LW1, 133; LW6, 99-11, LW16, 332). The DCC can offer a middle way between the philosophical and psychological notions of concept, in agreeing with the functional aspect of the psychological notion, while aiming to achieve the objectivity of the philosophical notion, through the DCC’s emphasis on action and public means and consequences. I now turn to clarifying the DCC and its benefits in CE over the psychological concept.

3. The Deweyan conception of concepts

On the DCC, concepts are abstractions from operations performed and, importantly, to be performed. An operation on some part of one’s environment can be specified in terms of the means-consequence relation. The means are the grounds on which one gets a handle on an entity, property, quality, or event, while the consequences are the ends towards which the means are put to work, the reaction of the means upon manipulation or operation. A concept, then, is a kind of rule for operations specified in terms of means and consequences. To make the DCC clearer and to demonstrate its import for CE, I will take a closer look at the function of concepts in the process of inquiry, as detailed by Dewey. This provides the grounds for thinking about the importance for concepts in structuring and directing our activities, situating concepts in the activity of inquiry.

In Dewey’s philosophical views, we can identify commitments to the process of inquiry as being reconstructive of a situation or environment and to inquiry itself being constituted by conceptual elements. It is emphasised that successful inquiries depend on a change in both ‘existential’ and ‘ideational’ materials (LW12, 121), meaning that inquiry involves the manipulation of concrete elements found in one’s environment and the manipulation of conceptual materials, or ideas.² For Dewey, successful inquiry is a matter of reconstructing an indeterminate situation into a determinate or understood situation. As Dewey put it,

Inquiry is the controlled or directed transformation of an indeterminate situation into one that is so determinate in its constituent distinctions and relations as to convert the elements of the original into a unified whole. (LW12, 108)

According to Dewey, this controlled or directed transformation of an indeterminate situation is not just a matter of changing one’s mind but also requires active intervention on and manipulation of the situation itself (LW12, 109–110) – successful inquiry depends on action, not just a change in mind. Active intervention and manipulation lead to reconstruction – or transformation – of the situation, opening new regularities that inquirers can exploit in anticipating and bringing about future experiences (LW4, 99–103; LW8, 205). As Dewey helpfully puts it, scientific operations ‘are such as disclose relationships’ (LW4, 100). These relationships, or relations, are an important part of what it means to make something meaningful; as Gronda (2020) makes clear, there is an inherent semantic component to the process of inquiry, where the articulation of new regularities that resolve inquiry is also a matter of conceptual articulation.³ Without intervention, these relations or patterns would be unknown, and the relations upon which concepts are built would remain hidden or, better yet, unarticulated.

As we can begin to discern, Dewey is committed to a broader conception of language and meaning than what is expressed in speech or writing alone. He thinks of language as being about ‘all kinds of signs and symbols’ (LW6, 4). Symbols, according to Dewey, are artificial notational devices used to communicate ideas, while signs are ‘natural’ and more direct, like smoke being a sign of fire (LW12, 57–58). Events can signify other events, implying that meaning extends beyond what is strictly linguistic; as in, meaning extends beyond what is expressed in speech and symbols alone. This, as noted above, is important, because he thinks about inquiry as being a matter of making situations – and phenomena, relations, and events therein – meaningful and so understood. The outcome of inquiry, in making phenomena, relations and events meaningful, is a matter of semantic (or conceptual) (re-)construction, as Gronda (2020, ch. 4) puts it. The bringing out of new relations, through manipulation, and so new means-consequence relations from which the situation can be made meaningful or better understood, is a way of (re-)constructing concepts from the materials of the problematic situation.

In inquiry, control is gained by using concepts, because of their meaning, articulating means-consequence relations that we can exploit. The meaning of a concept is given by the rule formulating the expectations of consequences by using certain means that arise in the activities and operations in which it figures. For example, the meaning of an everyday concept like ‘water’ is determined in relation to water being used as a means to consequences like thirst-quenching and washing; more complex means-consequence relations in rules as given by scientific concepts arise from activities like the boiling or freezing of water, which relates the concept ‘water’ to concepts like ‘vapour’ or ‘ice’ (cf. Gronda 2020, 82–84; Dewey LW4, 126–127). It is by virtue of operating on water in these ways that the concept ‘water’ achieves its meaning, in the means-consequence relations established in these activities. The entity, or object, water also becomes meaningful in this way (in a sense, it becomes water, since it is not individuated independently of these relations),⁴ as being seen as means to these consequences, while the concept becomes an abstraction into a rule based on these means-consequence relations. Thus, we can respond to problems in a controlled and rational way, by taking the qualities of water as means to particular consequences, a relation of means and consequences embodied in the rules expressed by the concept ‘water’.

According to Dewey, ‘what makes any proposition scientific is its power to yield understanding, insight, intellectual at-homeness, in connection with any existential state of affairs, by filling events with coherent and tested meanings’ (LW1, 129) (My emphasis). These coherent and tested meanings are ideas that are suggested in the process of inquiry, based on the ‘facts of the problematic situation’: they are ‘anticipated consequences (forecasts) of what will happen when certain operations are executed under and with respect to observed conditions’ (LW12, 113). Thinking of concepts in terms of expressing ‘anticipated consequences of what will happen when certain operations are’ performed helps us answer an important question about the function of concepts of inquiry, in how they participate in making a situation or an event understood by making it meaningful. Concepts play an important function in directing inquiry, in making action more than mere habitual response to stimuli – indeed, this is what Dewey emphasises as what traditional rationalists get right about concepts (LW4, 87–89). A Deweyan conception of the function of concepts will see them as rules for action and interpretation, for dealing with problematic situations, where, as noted above, a problematic situation is dealt with in transforming it into an understood or determinate situation. As Gronda helpfully puts things, Dewey views concepts as rules ‘for promoting certain controlled transformations of the environment so as to bring about certain expected consequences’ (Gronda 2020, 73).

The idea that the meaning of concepts is expressed in terms of rules can seem puzzling. However, it can be clarified by thinking more carefully about the role of ‘anticipated consequences’ that concepts articulate. The function of concepts is to direct or control inquiry, and they do so by playing a role in anticipating ‘consequences of what will happen when certain operations are performed’. According to Dewey, acting rationally or reasonably is a matter of applying concepts in this way, since ‘rationality is an affair of the relation of means and consequences’ (LW12, 17) (emphasis in original).⁵ This relation, as I see it, is key to understanding the connection between concepts, actions, and inquiry as a matter of controlled transformation of an environment. For Dewey, events are made meaningful, because of ‘the conversion of causal bonds, relations of succession, into a connection of means-consequence, into meanings’ – when cause–effect becomes means-consequence actors are ‘responding to things in their meanings’ (LW1, 277–278); ‘things gain meaning when they are used as means to bring about consequences (or as means to prevent the occurrence of undesired consequences), or as standing for consequences for which we have to discover means. The relation of means-consequence is the centre and heart of all understanding’ (LW8, 233). For Dewey, understanding is a matter of grasping meaning, and so, if the means-consequence relation is at the heart of understanding, it is also at the heart of meaning.⁶

This is because, when something is taken as a means – not as a mere habitual response to a cause – it is seen in a relationship to the consequences it brings about: it indicates what the consequences are of performing certain operations. The means-consequence relation spells out what it means for a concept to act as a rule for promoting certain controlled transformations, since it indicates what needs to be done in order to bring about some consequence. In this way, the means-consequence relation explains the way concepts are future-oriented – directions for action – since a rule promoting certain controlled transformations depends on the means for some consequences being the right means to the right consequences.⁷ Taking something as a means to some consequences is part of what makes a phenomenon meaningful or intelligible, in that it helps actors anticipate and control future experiences and events, anticipate problematic consequences, and respond to them in directed or controlled ways.⁸ This makes the DCC an appropriate basis for thinking about actionable CE, since CE is or should be a matter of taking control of our concepts, to take control of our lives and thoughts. This is in line with predecessors of CE, like Carnap’s explication. Richard Jeffrey, a student of Carnap, argues that Carnap’s ‘central idea was: ‘It’s high time we took charge of our own mental lives’ – time to engineer our own conceptual scheme (language, theories) as best we can to serve our own purposes’ (1992, 28)⁹ – time to take control of and with our concepts. We can understand an actionable account of CE based on the DCC and Dewey’s theory of inquiry as a basis for taking control, with these pragmatist tools in hand. I now return to showing how this conception of concepts, the DCC, can provide the basis for actionable CE, by showing its superiority over the philosophical and psychological concept, as presented in §1.

4. Engineering means-consequence relations

The problem, as we saw in §1, with the philosophical concept for CE is that it makes it puzzling how CE could be possible, since concepts are fixed by their extensions. The philosophical concept is too static to make CE possible. The benefit of the psychological concept is that, since it is based on the kind of functional-informational role that concepts play in people’s thoughts and activities, it makes CE possible and actionable. To be actionable, CE needs to be applicable to a variety of cases, and a view of concepts must be used to identify the ‘need for conceptual [engineering] and then implementing prescribed changes’ (Isaac 2021, §5.3). Thinking of concepts in terms of the DCC helps us achieve this, perhaps even better than what the psychological concept promises.

One clear benefit, immediately, is that concepts as rules for action provide a clear standard of successful function: a concept functions well if and only if it leads to effective action, in relation to some aim. The function of a concept is just that, to direct action and inquiry, on the DCC. Dewey would concur with contemporary conceptual engineers in emphasising that ‘[s]uccess is measured in efficacy’ (Nado 2020, 17) and that CE aims at ‘normatively ameliorat[ing] the cognitive quality of our conceptual devices in terms of their functional efficacy’ (Isaac 2021, §2). The DCC qualifies the efficacy of concepts as needing to have specific and determinate experiential and practical consequences, going back to Peirce’s (1878) pragmatic maxim. Since concepts play a role in determining actions, and concepts are revised on the basis of their successes and failures the DCC, with its account of concepts as rules that specify operations or actions to be performed in terms of the means-consequence relation, makes concepts a key part of our cognitive and practical lives. Concepts, according to Dewey, play an indispensable role in intelligent action, and so the DCC satisfies the criteria of scope and impact.

The DCC might even outperform the psychological concept in scope, since the psychological concept, as Isaac (2021, 5.1) formulates it, seems to primarily focus on the cognitive lives of people and the role that concepts play in thought. Isaac (2021) cites Machery’s (2017) discussion of the cognitive function of concepts approvingly: ‘Concepts determine the inferences we are prone to draw, and our thoughts follow their tracks’ (2017, 231). Isaac argues that the psychological concept can capture this well, since it emphasises the very role that concepts play in humans’ cognitive lives, whilst the philosophical concept does not capture this. The scope of the psychological notion is thus broader and more accommodating than the philosophical notion. However, as I see it, the DCC wins out over the psychological concept: concepts do not just guide thought through their determination of inferences, they also guide and determine action and make events and things meaningful. This can be best demonstrated by a brief comparison between the DCC and inferentialist conceptions of language and meaning, often inspired by Dewey’s views on language.

Recall that on the DCC, the meaning of concepts is given by the operations specified through rules stating means-consequence relations. The view that concepts gain meaning through the functional roles they play in inferences seems to capture a similar relation, but rather than this relation being a matter of means and consequences, it is a relation holding between premise and conclusion. Wilfrid Sellars’s (1969) functionalism and Robert Brandom’s (1998) thorough exposition of inferentialism captures this sentiment about the meanings of concepts. However, though both Sellars and Brandom place emphasis on the role of know-how in conceptual understanding, they focus on language alone: meaning is a linguistic phenomenon. For Dewey, meaning is also linguistic, in a specific sense of ‘linguistic’, but it goes beyond the use and utterance of words and sentences, to include taking things as signs (‘smoke is a sign of fire’), which stand in cause–effect relationships. These cause–effect relations can be made meaningful when we take the causes as means to specific effects as consequences which we anticipate and work for. As such, Dewey allows that things – actions, events, qualities, entities – gain meaning in their use; linguistic symbols are just another (albeit incredibly important) class of things that become meaningful through their use.¹⁰ Dewey thus provides a more general account of meaning than one restricted to inferences. To make this clear, we can see that premises are to conclusions what means are to consequences: you use the former to arrive at, or derive, the latter. Analogously to inferentialism, the focus on inferences and their role in our cognitive lives thus limits the psychological concept in a way that the DCC is not. For Dewey, the practical is primary, and thus we can ground the theoretical or cognitive in the various practical activities we engage in (LW4, chs. 5 and 6; 1916, ch. 14). This establishes a possible advantage of the DCC over the psychological concept of ‘concept’ in relation to the DCC making CE actionable in relation to the scope and impact of the DCC.

This, as I see it, is also in line with important characterisations of conceptual ethics and CE. For example, because the DCC emphasises changing actions, rather than just changing minds or inferences, it fits better with Burgess and Plunkett’s characterisation of the aims of CE and concepts:

our conceptual repertoire determines not only what we can think and say but also, as a result, what we can do and who we can be. In other words, which concepts we use has an important impact on the space of possible actions and lives available to us. (2013, 1091)

Additionally, as Kevin Scharp (2020, 410) puts it, ‘conceptual engineering is essentially oriented to action, not belief. Making changes to our conceptual scheme or our language is an activity.’ Not only is CE an activity but requires change in action for a change in concepts. Actionable CE is thus not just aimed at the psychological concept, but at the very actions and activities that constitute or uphold the concepts used, something the DCC allows for and encourages, given the relationship between means-consequence relations and actions.

Similarly, the DCC can provide a precise account of the basic constitutive features of concepts in terms of concepts being rules for operations to be performed as specified by means-consequence relations. Indeed, this kind of thinking is found in the close cousin of the DCC, operationalism, which is also motivated by the precisifying concepts for measurement and testing in terms of operations performed (Chang 2017; Vessonen 2021), being wary of applications of the concept that goes beyond those operations without establishing some kind of continuity between operations, while also aiming to extend those operations, making concepts more empirically meaningful through the new operations (Chang 2017). Similarly, the unambiguity of operationalist thinking, in terms of the theoretical fruitfulness it leads to is clear in science, where concepts can be understood through operations performed. This is particularly important in cases where scientists move beyond domains that are already well-understood, to domains where concepts might not clearly apply, unless operations that carry over the concept are found – a deep concern of Percy Bridgman in his work on pressure (Chang 2017). Thus, there is a clear case to be made for the unambiguity and precision of concepts on the DCC.^11,12

Finally, the DCC can provide clear operationalisible definitions of concepts in order to ameliorate them, in addition to rules for implementation of those concepts. This seems to be one of the main benefits of the DCC, since it – along with other pragmatist and operationalist accounts of concepts – is concerned with the specificity and clarity of the meaning of concepts in terms of their practical consequences, as well as being able to identify when a concept does not work as it should, when leading to unwanted or bad consequences in action. Indeed, given that the DCC can be seen as an explication and improvement of operationalist thinking (see footnote 11), we might think that applicability of operationalism to specific cases, especially in science, provides good grounds for thinking that the DCC provides grounds for making CE an actionable methodology.¹³

Here, I want to anticipate two objections, while also sharpening the reasons for moving to the DCC instead of the psychological concept. First, it is important to emphasise that the means-consequence relation is not the view that the ends alone justify the means, or only the consequences of actions matter independently of means. Brandom, for example, criticises classical pragmatists for only focusing on meaning-as-ends, ‘identifying propositional contents exclusively with the consequences of endorsing a claim’ (1998, 123), rather than taking into account the so-called circumstances of application for the claim as well as the consequences of endorsing it. However, Dewey’s sophisticated view of the relationship between means and consequences should allay any fear that the consequences, or ends, alone make up for the meanings of a concept or claim. In his (LW13), Dewey criticises the view that takes the ends alone to justify the means as the ‘fallacy involved in the position that ends have value independent of appraisal of means involved and independent of their own further causal efficacy’ (LW13, 229). Dewey thinks means and consequences are necessarily linked in a relation: a consequence is not a consequence without a means to it, and vice versa. This is because Dewey the relation ‘of means and consequence which defines an operation […] is a universal’ (LW4, 130). Dewey clarifies this idea of universals by speaking in terms of ‘if–then propositions’. A universal is here stated as a rule, ‘a formulation of an operation to be performed’ (LW11, 105, 107). Dewey thus formulates a functional account of universals as the expressions of concepts as rules for operations, in terms of the relation between means and consequences.¹⁴ Again, the operation to be performed needs to be understood through the aims of the operation itself and how to bring the consequences aimed at into existence. Unless we take the whole relation of means to consequence as the meaning of a concept, we lack the resources to anticipate future experiences: we are left without control of actions, since all we have will be either specific means, without relations to consequences, or consequences without means to achieve them.

Secondly, Cappelen (2018) objects to inferentialist thinking in CE on the grounds that it makes it difficult to understand how conceptual change and evolution takes place. As Cappelen puts it, inferentialism is ‘for the most part [a] static framework[…] and [it is] not constructed to account for constant evolution and revision’ (2018, 187). This, of course, is remedied by the psychological concept, since it looks at psychological-informational states of subjects, where conceptual change, revision, and evolution can be seen as a descriptive fact. However, the psychological concept faces a different objection by Cappelen (2018, 185) in discussing intentions behind use of concepts, raised as well by Woodward (2021, 20n9), that a psychologistic notion of concept is going to be too varied for the kinds of uses that we want to identify of concepts in CE. Indeed, the deeper point to be made here, in a Deweyan vein, is just that the concept of concept we want to use CE (and inquiry more generally, see §4) is not a psychological notion, but one that can better guide our actions. It is unclear how exactly the psychological concept can function normatively in this way. It is my contention that the DCC can balance between these problems, in that the appeal to operation in the context of activities is what gives meaning and stability while allowing for change in concepts. Stability is guaranteed in terms of agreement (or disagreement) between agent’s operations and activities – whether they use the same means to achieve the same ends, or not – while the possible changeability of the concept is guaranteed by the fact that operations and activities do change. However, often these changes are continuous, establishing a continuity between operations and activities, rather than breaking with previous usage completely. This makes the DCC a promising candidate for thinking about both conceptual change and conceptual revision, without the fickleness of the psychological concept or the rigidity of the philosophical concept.

5. Conceptual engineering as philosophical inquiry

Having established the DCC and explored its benefits over other conceptions of concepts in the previous two sections, I now turn to the final piece in my Deweyan conception of CE. We have already discussed the function of concepts in guiding inquiry, turning an indeterminate situation into one that is understood through the use and improvement of means-consequence relations. This is an active process, with a view of concepts specifically as bottoming out in operations to be performed, suggesting an active notion of CE, focusing on performing and changing activities to change concepts and vice versa. But Dewey has a specific idea of how inquiry proceeds. His model of inquiry provides five phases of inquiry, leading from an indeterminate situation to settled judgement. By making use of this five-stage model of inquiry, I think we can get clearer on how CE might proceed, while also allaying fears about straying too much from the spirit of Dewey’s thoroughgoing naturalism. This five-stage pattern of inquiry is also an improvement on current accounts of the process of CE that focus on either two or three phases of engineering, as I will discuss more below.

At the outset, it is worth clarifying that Dewey does not, as far as I am aware, apply his theory of inquiry to philosophy itself. The theory of inquiry that Dewey supplies in (LW8) and (LW12) is primarily concerned with the structure of scientific inquiry and experimental learning. Dewey does, however, provide a normative role for philosophy to play through criticism. This view of philosophical methodology is a good way into thinking about the theory of inquiry when focused on concepts, as fulfilling the role of philosophy as criticism. Dewey argued that philosophy is

criticism which traces the beliefs to their generating conditions as far as may be, which tracks them to their results, which considers the mutual compatibility of the elements of the total structure of beliefs. Such an examination terminates […] in a projection of them into a new perspective which leads to new surveys of possibilities. (LW6, 19) (My emphases.)

The materials of philosophy are derived from previous inquiries, what scientific practice has related back to common sense and ‘culture’, or experience (LW12, ch. 4).¹⁵ Philosophy thus plays a critical-normative role, while also paying close attention to the origins and uses of beliefs or concepts. This is important for Dewey, because he is committed to naturalism about philosophy: philosophical methodology and subject-matter is continuous with, and does not go beyond, the methods and subject-matters of science or other (practical) arts (LW1, ch. 11).

Some CEs do take a two or three-stage approach to the process of engineering, like Chalmers (2020), who takes CE to consist in three stages: the design stage (DS), the implementation stage (IS), and the evaluation stage (ES). DS is a matter of designing a concept. IS is a matter of making use of the newly designed concept, perhaps engaging in conceptual activism (Cappelen 2018) to get others to use it. Finally, ES is the stage of assessing whether the concept does a good job in the role it has been put into.

These three stages seem to capture only the latter half of a full process of engineering a concept. Designing a (new) concept requires some purpose to which the design is for, which requires establishing some context and problem which the concept is meant to be used for. Even if engineering is de novo, in that it is a matter of creating a new concept, this concept still has to be suitable to some problem or context of use. This requires clarity on the problem and context of use. A three-stage process like the one Chalmers describes here leaves this preceding stage of assessment of the problem and context untouched (see also Cappelen 2018). I think we can extract a much richer account of the process of engineering from Dewey’s extensive writing on the pattern of inquiry. This helps us put the preceding discussion of concepts as rules for operations specified through means-consequence relation into a philosophical framework of CE as philosophical inquiry. We can reorientate Dewey’s account of inquiry to focus on concepts, rather than whole situations (though, of course, a modification of concepts will also involve a modification of a situation). Philosophical inquiry will be seen as a matter of evaluating and transforming defective concepts for various purposes for which we need the concepts, along the lines of how Scharp (2020) suggests philosophy is the study of defective concepts.

Dewey suggests that there are five stages in the pattern of inquiry, between moving from an indeterminate situation – through inquiry – to a warranted judgement, or a determinate situation.¹⁶ The five stages, as Matthew Brown (2012) discusses them, are

1. Observation collects data or determines the facts of the case in order to take stock of the fixed conditions of the situation that shape the problem.

2. Institution of a problem specifies what is perplexing or indeterminate in the situation in order to formulate a problem statement.

3. Suggestion proposes hypotheses (or ideas) for solving the problem.

4. Reasoning refines hypotheses, brings hypotheses into relation with larger theories and conceptual frameworks, coordinates observations and hypothesis, suggests new observations, and proposes experimental tests.

5. Experimentation tests the hypothesis by tentative or limited application to the situation that involves intervening or acting in the situation. (2012, 286)

This, I think, describes a fruitful way of conceiving of the process of CE as well, as not just experimental inquiry.¹⁷ Instead of thinking of the process of engineering in two or three steps – evaluation, improvement, implementation – we disambiguate the various tasks that need to be performed in the steps described broadly as evaluation and improvement in terms of the ideas of observation and institution of a problem, and suggestion and reasoning, respectively, and finally experimentation.

In the first phase of philosophical inquiry, to co-opt Dewey’s pattern of inquiry, there will be the observation of the relevant data, purposes and aims of concepts, and – since a key component of concepts is their guidance of action – the role they play or function they have in agents’ lives. Determining what a concept is will therefore also be a matter of figuring out who the concept is used by and what the use of the concept has been.¹⁸ Similarly, data, as Dewey puts it, are taken, and so whatever the concept that needs to be engineered is, it will have to be clarified in some way, just like data and the facts of the case are in Dewey’s process of inquiry.¹⁹ Various methods can be used in this case, and recent discussions of the complementarity of conceptual genealogy and CE provides a useful example of how the case of observation in philosophical inquiry might proceed. Genealogies can help clarify the function and history of use of the concept to be engineered, making clear what the potential deficiencies and positive functions of a concept might be (Koopman 2011; Reck 2012; Dutilh Novaes 2020; Queloz 2021). On the DCC detailed above, the work done here is a matter of identifying the means and consequences specified by concepts as rules for operations to be performed. As Dewey put it above, this stage traces concepts to ‘their generating conditions as far as may be, which tracks them to their results’ (LW6, 19).

Importantly, the kind of empirical psychological research Isaac (2021) draws on to argue for the fruitfulness of the psychological concept can be utilised in this stage of inquiry, to identify potential means to engineer.²⁰ This stage, since it lays the foundations for what is to come further in inquiry, can afford to be opportunistic about the methods employed; inquiry does not predetermine what methods will lead to fruitful consequences, and so we should let a thousand flowers bloom at this stage, employing the various methods philosophers have suggested in studying concepts, from conceptual analysis to empirical psychology and experimental philosophy (cf. Machery 2017 on experimental philosophy and rebooted conceptual analysis). The main point to take away, however, is to recognise that this is the beginning of inquiry, and not its end. The subsequent problematisation of the data and the engineering of the concepts is to come.

Secondly, having clarified what the means and consequences to be engineered are, what the facts of the case are – as in, the function of the concept – and who the relevant actors applying the concept is, the phase of institution of a problem in the case of philosophical inquiry, is a matter of establishing what the purported deficiencies of a concept is. As in the case of inquiry, what the problem is (and what might serve as possible solutions to the problem), will depend on the specific facts of the case. Importantly, it will depend on how the means we thought were means to our expected consequences failed, thus leading to a failure of the concept in guiding action (at least in some context). Broadly speaking, this will imply determining along which lines the concept is meant to function and how it falls short of the purported function. Concepts can be deficient in many different ways, ranging from ethical and political deficiencies, to logical and theoretical deficiencies. How deficient a concept is, what kind of deficiency it suffers from, and how it needs to be modified will thus depend on the purpose of – the consequences wanted from – the use of the concept and who the concept is used by. For example, Scharp (2013) argues that the concept of truth is deficient in that it is logically inconsistent, i.e. the constitutive principles of the concept leads to paradox. This a very serious deficiency for some purposes and for the use of the concept in relation to some aims. As Scharp emphasises, the concept of truth should be replaced for certain theoretical purposes, while the concept might be totally fine in ordinary contexts. The formulation of the problem thus needs to be sensitive to the facts of the case established in the first phase of philosophical inquiry, in order to put the subsequent engineering of the concept on the right track, to identify the right means and consequences that will resolve the deficiency.

These two first phases of philosophical inquiry leave room for the use of philosophical methods like genealogy, analysis, ordinary language philosophy, psychology, and experimental philosophy. The use of method will depend on the problem at hand: in a Deweyan spirit, there is no one-size fits all judgement to be made ahead of inquiry, but the important point to note about these methodologies is that they are not the final say in philosophical inquiry. They provide the grounds on which we perform the subsequent reconstruction and engineering.

The third stage of philosophical inquiry is the beginning of the revisionary element of CE. Having clarified the facts of the case (i.e. the concept to be engineered) and instituted a problem (i.e. made clear the deficiencies of the concept), we proceed to the possible suggestions for a solution to the problem. Here, conceptual engineers need to pay close attention to the aims of engineering the concepts, while suggesting potential revisions of the concept that manages to capture the functions of the original concept we thought were worth keeping as well as avoiding its deficiencies – this means delimiting the relevant operations for the purposes of some activity, while trying eliminate those that fail, aiming to secure a continuity in operations and activities, while also modifying other activities.²¹ As with operationalism, the aim of revision and conceptual extension on this picture is made possible by paying close attention to and ‘crafting of more uses of [concepts] in harmony with each other’ (Chang 2017, 36). Potential suggestions of concepts will be along the lines of suggesting new means to consequences or aims we wish to achieve, or potentially even a matter of changing our aims and so the consequences we strive towards. Either way, this requires working out what kinds of means-consequence relations will lead to successful action, since the function of a concept – and a successful revision of concepts – is a matter of providing rules for operations to be performed.

The fourth step is a matter of reasoning out the possible working of the suggested concept by relating them to other concepts, practices, and philosophical accounts, before implementing them in new activities. Since concepts are guides for action, it can be difficult and risky to implement them directly without reasoning about their possible consequences beforehand. The stage of reasoning brings the concept into connection with other concepts, enriching its meaning and significance (as Chang 2017, 36 puts it, ‘[c]oncepts become more meaningful if they exist within a more thickly connected set of practices’²²). If the consequences do not eliminate the deficiencies of the concept we originally wished to change, we step back to one of the earlier stages of philosophical inquiry. Something may have gone wrong in any of the stages, in establishing the facts of the case, formulating the deficiencies of the concept, or in the suggestions we have come up with for revision.²³ If we think the consequences of the proposed concept are positive, in that they look to resolve the deficiencies identified in the first and second stages of inquiry, then we proceed through to the fifth stage.

Finally, the concept is tested through experimentation. In the case of many philosophical concepts, dealing with abstract notions of knowledge, reality, or truth, the experimental element of philosophical inquiry might be best understood as the implementation of the concept in the various practices it is relevant to. For Dewey there is a real sense in which the consequences of the new concepts need to go beyond merely philosophical problems. Hence his suggested reconstructions – or revised concepts – of ‘knowledge’, ‘truth’, ‘reality’, and ‘the good’, all of which he relates to particular activities, or practical engagements with experience.²⁴ In the case of practical, ethical, or political concepts, experimentation and implementation will be two sides of the same coin. For Dewey, ethical and political goods were subject to revision in much the same way that scientific concepts were (MW12, chs. 7 and 8). Even in relation to moral and political concepts, there is an experimental element, in that the implementation of new concepts – new means-consequence relations – can only be established based on whether they help us achieve our aims in the actions that they facilitate. The concept, to use Carnap’s words, proves itself by way of its fruitfulness, through the process of inquiry as described here.²⁵

On this picture of philosophical inquiry, CE becomes a matter of actively taking control of concepts, by changing our actions in accordance with the means that bring about the wanted and anticipated consequences, or by establishing new means-consequence relations, that best fit with the activities we want to perform. The DCC, with its emphasis on the means-consequence relation, thus provides a grounded approach to actionable CE, by looking at CE as a matter of changing actions, not just of changing minds.

6. Conclusion: prospects for active conceptual engineering

This concludes the sketch of a pragmatist account of CE in line with Dewey’s accounts of concepts and inquiry. There is much more to be said on this front, especially on the ways in which deficient concepts may be discovered, to be engineered. Because, as Dewey (LW8, 199) notes, philosophical problems may be far removed from particular data, philosophers as conceptual engineers should think of themselves as trying to identify and, in some sense, proliferate deficiencies in concepts in establishing their use in concrete practices. If CE can be envisaged as a methodology that aims to really bring out the deficiencies of concepts, and not merely resolve them, we get what we might call an active conception of conceptual engineering that fits well with Dewey’s theory of inquiry.²⁶ CE should therefore be a commitment to testing concepts to their limits, to see where they break down, in order to actively modify them, instead of sitting back and waiting for problems to arise. This requires a thoroughgoing active and critical attitude. To use Scharp’s (2020) characterisation of CE, we see that on this active conception, philosophy truly becomes the study of defective concepts through activity, in actively trying to identify and bring those deficiencies to the fore and revise concepts in light of those deficiencies. We can think about this as analogous to scientists experimenting to bring about new problems, rather than simply sitting back, waiting for them to arise. If this picture of philosophical inquiry as active CE gains traction, we might see the ensuing reconstruction of philosophy that Dewey (MW12) advocated, since our philosophical concepts would also need to be put in contact with concrete practices and practical experience.

This paper has provided the foundations for this conception of CE, by formulating and arguing for the superiority of the DCC, combining it with Dewey’s view of the structure of inquiry, providing a full picture of how the DCC might be put to work, while also disambiguating the different stages of CE. This work demonstrates that Dewey’s philosophical work has continued value, especially in relation to the CE literature, given his focus on concepts as bearing out their significance in action through means-consequence relations, his theory of inquiry, and his overall reconstructive pragmatism, leaving us with a Deweyan conception of philosophical inquiry through the conceptual engineering of means-consequence relations.

Acknowledgements

I’d like to thank Céline Henne for detailed discussions and feedback on earlier versions and ideas of this paper, as well as Yvonne Hütter-Almerigi and an anonymous reviewer for helpful feedback. Thanks as well to Hasok Chang, Marabel Riesmeier, Rory Kent, Jouni-Matti Kuukkanen, Miguel Ohnesorge, and the audience at the Conceptual Engineering and Pragmatism Online Workshop for commenting on a previous draft of this paper.

Disclosure statement

No potential conflict of interest was reported by the authors.

Additional information

Funding

This work was supported by Cambridge Commonwealth, European and International Trust [Grant Number 10511726].

Notes

1 We might worry about the level of explanation of the psychological concept, in that it is not a given that it provides a functional description of concept, depending on what we mean by ‘functional’. There is a real question about whether the psychological notion aims at manifest or operative concepts (Haslanger 2006), and another question in addition about how that matches with the target concept in an ameliorative project.

2 Dewey uses many different notions to discuss what we would call concepts: ‘meanings’, ‘ideas’, ‘conceptions’, and only sometimes the notion ‘concept’. See Gronda (2020, 50) for discussion.

3 I thank an anonymous referee for pushing the point of whether (a) concepts are tools for the resolution of an indeterminate situation or (b) concepts are the subject or object of transformative inquiry. As I see Gronda’s semantic point, in Dewey’s account of inquiry, (a) and (b) are necessarily related.

4 I thank Céline Henne for helping me clarify this point.

5 See also (LW13, 57): ‘Intelligent activity is distinguished from aimless activity by the fact that it involves selection of means – analysis – out of the variety of conditions that are present, and their arrangement – synthesis – to reach an intended aim or purpose.’

6 ' To understand is to grasp meaning.’ (LW8, 221) See also (LW13, 56): ‘Growth in judgment and understanding is essentially growth in ability to form purposes and to select and arrange means for their realization.’

7 After all, ‘action is at the heart of ideas’ since ‘[i]deas direct operations’ (LW4, 134)

8 Fesmire (2015, 72–73) makes a similar point, though does not spend much time on relating the means-consequence relation to meanings or concepts.

9 See Dutilh Novaes (2020) for discussion, where I first found this quote by Jeffrey.

10 Dewey (LW8, 231) is particularly clear on things being meaningful, as opposed to just sentences and words.

11 Dewey’s view of meaning has a lot in common with current views of operationalism, as Gronda (2020, 73–75, 85). A key difference between Bridgman’s (1927) operationalism and Dewey’s pragmatism is the latter’s emphasis on operations bringing out means-consequence relations, as has been emphasised here, while Bridgman focuses on the operations alone. I see the DCC as an explication of operationalism, given Dewey’s (LW4, ch. 4–5) alliance with operationalism. As such, I think any benefit operationalism enjoys can be taken up by the DCC.

12 Precision does not imply exactness. Vague concepts can be more useful for specific purposes than an exact one, and so more precise in the above meaning. I think this picture of CE and concepts is able to accommodate this point.

13 Here I am thinking of Bridgman’s (1927) discussion of various scientific concepts, like ‘space’, ‘time’, etc. See Chang (2017) for elaboration and discussion.

14 In traditional terms, Dewey also identifies the ‘conceptual contents which anticipate a possible solution and which direct observational operations’ with predicates (LW12, 127–129).

15 As he puts it in (LW1, 305), philosophy’s ‘business is to accept and to utilize for a purpose the best available knowledge of its own time and place. And this purpose is criticism of beliefs, institutions, customs, policies with respect to their bearing upon good.’

16 See Dewey (LW8, 200–207) and (LW12, ch. 6). Brown (2012) helpfully distinguishes between the temporal sequence of inquiry and the functional pattern of inquiry. See fig. 1.

17 Isaac (in progress) also formulates a five-stage account (the DEDTI-model) of CE.

18 We might think of this along the lines of Queloz’s (2021) thinking about the practical origins of ideas, or Hannon’s (2019) function-first epistemology. Both views have a lot in common with Austin’s (1979, 182) discussion of the underlying rationales of our ‘common stock of words’.

19 Dewey (LW8, 199) remarks upon the difficulty of identifying data in philosophical problems.

20 It can be used to identify the manifest or operative concept(s), for example. See footnote 1.

21 We want to do responsible CE, as Queloz (2021, 42) puts it, paying attention to both positive and negative aspects of a concept. Engineering is not just about paying attention to defects, but also their good or useful functions. See also Chang (2017, 35–36) on multiplicity and unity in operationalism.

22 Chang (2017, 36) continues in a Deweyan spirit: ‘more meaningful concepts are more effective facilitators of inquiry about nature.’ Dewey also thinks that our ability to understand, our ability to achieve the aims of inquiry, ‘is immensely furthered by language and by elaboration of a series of meanings and through reasoning’ (LW8, 232).

23 As Brown (2012) emphasises, the process of inquiry, on Dewey’s view, is not totally linear.

24 See (MW12) for discussion of all concepts, (LW1) for discussion of reality, and (LW4, LW8, LW12) for discussion of knowledge. Dewey’s (1916) also provides extensive discussion of truth and knowledge.

25 See also Dewey (LW8, 266) on application of concepts to new situations, which would satisfy another notion of fruitfulness from Stuart (2016).

26 Active CE is inspired by Chang’s (2012) active realism and active pluralism, which encourages actively pursuing resistance with reality through the proliferation of systems of practice in science. Chang (2017) puts operationalism to work in this vein, arguing for a multiplicity of measurement methods and concepts.