Article
Author: Jonathan Cottrell (University of Edinburgh)
Some of Hume’s central arguments in the Treatise—for example, arguments about causality, the self, and motivation—concern which of our perceptions represent, and what these perceptions can and cannot represent. A growing body of literature aims to reconstruct the theory of mental representation that (it is presumed) underwrites these arguments. The most popular type of interpretation says that, according to Hume’s theory, copying plays a significant role in explaining mental representation. This paper raises two challenges to such interpretations. First, they cannot be squared with Treatise Book 2’s account of passions formed via sympathy (hereinafter, sympathetic passions). Second, Hume’s treatment of copying and representation in his later works differs significantly from that of the Treatise, and provides no clear evidence that the mature Hume accepted a theory of mental representation based on copying.
Keywords:
How to Cite: Cottrell, J. (2021) “Representation and Copying in Hume’s Treatise and Later Works”, Ergo an Open Access Journal of Philosophy. 8(0). doi: https://doi.org/10.3998/ergo.1141
Some of Hume’s central arguments in the Treatise—for example, arguments about causality (T 1.3.14.11/160–61),1 the self (T 1.4.6.2/251–52),2 and motivation (T 2.3.3.5/415, 3.1.1.9/458)—concern which of our perceptions represent, and what these perceptions can and cannot represent. A growing body of literature aims to reconstruct the theory of mental representation that (it is presumed) underwrites these arguments.3 The most popular type of interpretation says that, according to Hume’s theory, copying plays a significant role in explaining mental representation. This paper raises two challenges to such interpretations. First, they cannot be squared with Treatise Book 2’s account of passions formed via sympathy (hereinafter, sympathetic passions). Second, Hume’s treatment of copying and representation in his later works differs significantly from that of the Treatise, and provides no clear evidence that the mature Hume accepted a theory of mental representation based on copying.
I proceed as follows. Section 1 explains what is meant by ‘theory of mental representation’. Section 2 presents the view that Hume explains mental representation in terms of copying, along with some of its main textual support. Sections 3 and 4 present the challenge based on sympathetic passions. Section 5 considers two non-standard interpretations of Hume on representation and copying due to Karl Schafer (2015) and Donald Ainslie (2015; 2019), and argues that they face similar challenges. Section 6 presents the challenge based on Hume’s later works.
What is a theory of mental representation? According to an influential paper by Don Garrett, such a theory aims to “account for the intentionality of mental states—that is, their ability to represent or otherwise to be of other things”4 (Garrett 2006: 302). For Hume, having an intentional mental state that is directed upon a certain object (for example, believing in God) or that has a certain propositional content (for example, believing that Caesar died in his bed) constitutively involves5 having, in one’s mind, a mental particular—a “perception”—that has the relevant object or proposition as its “representational content,” to borrow a phrase from Garrett (2006: 302).6 Therefore, for Hume, the task of “account[ing] for the intentionality of mental states” becomes that of explaining the representational properties of perceptions. This task has two components:7 to explain both (1) “why … perceptions represent—that is, what makes them representational at all,” and (2) “how … perceptions represent—that is, what determines their specific representational content” (Garrett 2006: 302). A Humean theory of mental representation, as I understand it, would systematically answer these two explanatory demands.8 Its answers would provide a basis for Hume’s claims about which perceptions represent, and about what kinds of things our perceptions can and cannot represent.
In the recent Hume literature, most scholars agree that his answers to these explanatory demands will be naturalistic. That is, they will not posit any “explanatorily basic intentional properties” (Garrett 2006: 302). Hume will explain the representationality of perceptions, and the specific contents of representational perceptions, in terms of properties and relations that are not, themselves, intentional—in other words, properties and relations that could, individually, belong to something without intentionality. (Ainslie’s interpretation is non-naturalistic; see Section 5, below.)
In Treatise Book 1’s opening section, Hume argues for his Copy Principle: “all our simple ideas in their first appearance are deriv’d from simple impressions, which are correspondent to them, and which they exactly represent” (T 1.1.1.7/4)—or, as he says later, “all our ideas are copy’d from our impressions” (T 1.1.3.4/10; see also T 1.3.1.7/72, 1.3.7.5/96, and 1.3.14.16/163). Scholars often regard this principle as the core of his theory of mental representation. For example, David Landy (2018: 333) writes: “the Copy Principle, properly emended, constitutes the entirety of Hume’s theory of mental representation.” Landy’s emendations to the Copy Principle concern complex perceptions, which the principle does not mention (Landy 2018: 335–37). For present purposes, we can afford to focus on simple perceptions; we can therefore set these emendations aside. According to Landy and others,9 Hume holds that copying—understood as a combination of causality and close resemblance10—explains the representational character of simple perceptions as follows: a simple perception is representational iff it is a copy; and when a simple perception is representational, it represents that and only that from which it is copied.11 Let us call this theory the Copy Theory of Representation (hereinafter, CTR).12 Note that CTR is naturalistic: it explains the intentionality of mental states by appealing to a combination of resemblance and causation among perceptions—relations that are, individually, instantiated by things without intentionality.
Numerous passages may be cited as evidence that Hume accepts CTR. When taken together, the early passage introducing the Copy Principle and the later passage containing the “Representation Argument”13 give evidence that he accepts CTR’s claim that a simple perception is representational iff it is a copy. The former gives evidence that he accepts one direction of this conditional: a simple perception is representational if it is a copy. The Copy Principle asserts that every simple idea is caused by (“deriv’d from”), resembles (is “correspondent to”),14 and is a representation of (“exactly represents”) a simple impression. Hume claims to “establish” the Copy Principle in this section of the Treatise (T 1.1.1.12/7). But his explicit arguments for the principle concern only causation and resemblance. First, Hume argues that every simple idea exactly resembles a simple impression, save for its degree of liveliness (T 1.1.1.3–6/2–4).15 Second, he argues that every simple idea is caused by its resembling or “correspondent” impression (T 1.1.1.8–9/4–5). Taken together, these results imply that every simple idea is copied from a simple impression. In claiming to have “establish[ed]” the Copy Principle on the basis of these two arguments, Hume seems to assume that a simple perception is representational (“exactly represents”) if it is a copy.
The Representation Argument, which concerns passions, is often taken as evidence that Hume accepts the other direction of the conditional above: a simple perception is representational only if it is a copy. In the Treatise, Hume regards passions as simple impressions:16
The passions of pride and humility being simple and uniform impressions, ‘tis impossible we can ever, by a multitude of words, give a just definition of them, or indeed of any of the passions. (T 2.1.2.1/277, emphasis added; see also 2.2.1.1/329)
Concerning these simple perceptions, the Representation Argument goes as follows:
A passion is an original existence, or, if you will, modification of existence, and contains not any representative quality, which renders it a copy of any other existence or modification. When I am angry, I am actually possest with the passion, and in that emotion have no more a reference to any other object, than when I am thirsty, or sick, or more than five foot high. ‘Tis impossible, therefore, that this passion can be oppos’d by, or be contradictory to truth and reason; since this contradiction consists in the disagreement of ideas, consider’d as copies, with those objects, which they represent. (T 2.3.3.5/415)
In the course of arguing that passions can be neither “oppos’d by” nor “contradictory to” reason, Hume claims that passions are non-representational: a passion “contains not any representative quality”; feeling a passion involves “no more a reference to any other object” than does being sick, or being more than five feet tall.17 Scholars who cite this passage as evidence that Hume accepts CTR regard this claim as a lemma supported by the following argument: a passion is not copied from anything (it is “an original existence”—an archetype, not an ectype); therefore, a passion is not representational. So interpreted, the Representation Argument uses, as a suppressed premise, CTR’s claim that a simple perception is representational only if it is a copy. (Some scholars question whether Hume should have made this argument, given his other views about the passions; for discussion, see Section 4.)
There is also evidence that Hume accepts CTR’s claim that, when a simple perception is representational, it represents that and only that from which it is copied. Several passages suggest that, when a simple perception is representational, it represents that from which it is copied. These include the passage that introduces the Copy Principle, where Hume assumes that, if a simple perception is a copy, then it is representational and, moreover, represents the particular simple impression from which it is copied. They also include Hume’s claim that ideas “always represent the objects or impressions, from which they are deriv’d” or copied (T 1.2.3.11/37; see also 1.3.14.6/157, 1.3.14.11/161).
Other passages suggest that, when a simple perception is representational, it represents only that from which it is copied. Hume often seems to assume that an idea represents an individual of a certain kind only if copied from one. For example, he argues: “All ideas are deriv’d from, and represent impressions. We never have any impression, that contains any power or efficacy. We never therefore have any idea of power” (T 1.3.14.11/161). On one seemingly plausible interpretation, Hume means: we have no impression that is, or has a part that is, an instance of causal power; so, we have no idea copied (“deriv’d”) from an instance of power; and so, we have no idea that represents an instance of power.18 So interpreted, he seems to assume that an idea represents an instance of power only if copied from one. Hume gives similar arguments to show that we have no ideas representing external objects “specifically different” from perceptions (T 1.2.6.8/67–68) or a self with perfect identity and simplicity (T 1.4.6.1–2/251–52). Taken together, these passages give evidence that, for Hume, when a perception is representational (a fortiori, when a simple perception is representational), it represents only that from which it is copied.
So, the text of the Treatise provides significant evidence that Hume accepts the claims that comprise CTR: a simple perception is representational iff it is a copy; and when a simple perception is representational, it represents that and only that from which it is copied.
In the literature to date, we find two main objections to this kind of interpretation. First, some scholars19 object that CTR conflicts with the Treatise section “Of abstract ideas” (T 1.1.7), which seems to imply that two simple ideas copied from the same thing sometimes differ in what they represent.20 Suppose that two simple ideas, copied from the same simple impression of a blue point, stand in different associations: one is associated with the term ‘blue’, which is in turn associated with all and only one’s other ideas of blue individuals; the other is associated with the term ‘point’, which is in turn associated with all and only one’s other ideas of spatial points. Hume’s theory seems to imply that the first idea serves as a general idea that represents all and only blue things, while the second serves as a general idea that represents all and only spatial points, hence that the two ideas differ in what they represent—for some blue things are not points, and some points are not blue (T 1.1.7.7–10/20–22 and T 1.1.7.7n5App/637). This result conflicts with CTR, which implies that—since they are copied from the same thing—these two simple ideas are entirely alike in what they represent.
Second, some scholars21 object that CTR conflicts with Hume’s views about impressions of sensation.22 CTR implies that, if a simple impression of sensation is representational, then it is a copy, hence resembles its cause. But Hume seems to hold that (at least some) impressions of sensation are representational, while remaining completely agnostic about the properties of their causes. He suggests that some impressions of sensation are representational when he writes that “our senses … represent as minute and uncompounded what is really great and compos’d of a vast number of parts” (T 1.2.1.5/28) and seems to say that at least one such impression is representational when he writes of “that compound impression, which represents extension” (T 1.2.3.15/38).23 He expresses complete agnosticism about the properties of such impressions’ causes when he writes:
As to those impressions, which arise from the senses, their ultimate cause is, in my opinion, perfectly inexplicable by human reason, and ‘twill always be impossible to decide with certainty, whether they arise immediately from the object, or are produc’d by the creative power of the mind, or are deriv’d from the author of our being. Nor is such a question any way material to our present purpose. We may draw inferences from the coherence of our perceptions, whether they be true or false; whether they represent nature justly, or be mere illusions of the senses. (T 1.3.5.2/84)
Suppose, for the sake of argument, that if some impressions of sensation are representational, then some simple impressions of sensation are representational. If Hume accepted CTR, then he could not consistently hold that a simple impression is representational while remaining agnostic about whether it resembles its cause. But in the passages just quoted, when taken together, he seems to do just that. Therefore, these passages may be taken, collectively, as evidence that he does not accept CTR.
It is controversial whether either objection succeeds.24 I do not wish to adjudicate this issue here. Instead, I will raise a new challenge to the view that Hume accepts CTR, focusing on his account of sympathetic passions.
To “sympathize,” in Hume’s sense, is “to receive by communication” another person’s “inclinations and sentiments,” which include “opinion[s]” and “passions” (T 2.1.11.2/316–17). Let us focus on the latter. We have seen that, in the Representation Argument (T 2.3.3.5/415), Hume claims that passions are not representational. I will now argue that, if Hume accepts CTR, then he is committed to holding that sympathetic passions are copied from past perceptions. So, he cannot consistently accept CTR’s claim that a simple perception is representational if it is a copy.
As described in the Treatise section “Of the love of fame” (T 2.1.11), the process of sympathetic “communication” works as follows.25 Observing the outward signs of a certain passion in another person’s behaviour conveys to us an idea of that passion (T 2.1.11.3/317).26 Our ideas of other people’s passions are typically associated with our “idea, or rather impression of ourselves,” which is a very lively perception (T 2.1.11.4/317). Liveliness is transmitted, via this associative link, from our perception of self to our idea of the other person’s passion. Consequently, this idea is “converted into” an instance of the passion that it represents (T 2.1.11.3/317, 2.1.11.7/319, 2.1.11.8/319).
For Hume, then, sympathizing begins with an idea of another person’s passion; let us call it the initiating idea. We have seen that Hume regards passions as simple impressions (T 2.1.2.1/277, 2.2.1.1/329). Presumably, then, the initiating idea—which is “of” (T 2.1.11.3/317) or “represent[s]” (T 2.1.11.8/319) a passion— is itself a simple idea. (Or, at least, in some cases of sympathy, one’s idea of the other person’s passion is simple; let us focus on such a case.)27 Therefore, by the Copy Principle, the initiating idea one forms, when sympathizing with another person’s passion, is copied from a past token of that type of passion.
When one sympathizes, this initiating idea is “converted into” a sympathetic passion. Hume’s talk of “conversion” can be understood in two different ways. On one interpretation, Hume means that the initiating idea becomes the sympathetic passion. He seems to take this view when he writes that this idea “acquires such a degree of force and vivacity, as to become the very passion itself” (T 2.1.11.3/317) and that “an idea of a sentiment or passion, may by this means be so enliven’d as to become the very sentiment or passion” (T 2.1.11.7/319).
If this first interpretation is correct, there is one perception that is first an idea (the initiating idea) and later a passion (the sympathetic passion). When an idea, it is a copy of an earlier passion. In the process of sympathetic communication, it gains liveliness from another perception: the “idea, or rather impression of ourselves.” Hume mentions no other change on the initiating idea during this process. If he accepts CTR, then he cannot consistently say that a simple idea copied from one perception ceases to be a copy just in virtue of its receiving liveliness from another perception (I will argue this point in a moment). So, if he accepts CTR, he must (on pain of inconsistency) accept that the perception that is first an initiating idea remains a copy, even after being enlivened so as to become a sympathetic passion.
Why can’t Hume—assuming he accepts CTR—consistently say that a simple idea copied from one perception ceases to be a copy just in virtue of its receiving liveliness from another perception? Doing so would conflict with his account of beliefs formed through probable reasoning, which is structurally parallel to his account of sympathy (T 2.1.11.8/319–20). I shall explain. Let ‘A’ and ‘B’ denote types of objects or events whose tokens can be represented by simple perceptions. Suppose that, in Smith’s past experience, every A has been followed by a B. Smith wonders whether a B will occur, entertaining non-lively simple ideas of a B’s occurring and of no B’s occurring. She then observes (has an impression of) an A. By probable reasoning, she forms the belief that a B will occur. According to Hume, Smith’s forming this belief consists in the enlivening of a simple idea—her simple idea of a B’s occurring—by her impression of an A (T 1.3.8.7–12/101–3). If CTR is true, then this case involves a simple idea that is copied from one perception but enlivened by another: by CTR, Smith’s simple idea of a B is copied from a past simple impression of a B, but is enlivened by another perception—a present impression of an A. Suppose Hume said that a simple idea copied from one perception ceases to be a copy just in virtue of its receiving liveliness from another. He must then accept that Smith’s belief that a B will occur is not a copy. By CTR, it follows that this belief (since, by hypothesis, it is a lively simple idea) is not representational. But this result is unacceptable: beliefs are ideas (T 1.3.7.5/96), and every Humean idea is representational (T 1.2.3.11/37, 1.3.14.6/157). So, if Hume accepts CTR, he cannot consistently say that a simple idea copied from one perception ceases to be a copy just in virtue of its receiving liveliness from another.
So, if our first interpretation of sympathetic “conversion” is correct—that is, if the initiating idea becomes the sympathetic passion—and Hume accepts CTR, then he must (for consistency) accept that sympathetic passions are copies. Note that, in arguing this point, I have not presupposed any specific view of what copying is.
On the second interpretation of “conversion,” Hume means that the initiating idea is succeeded by the sympathetic passion.28 So no one perception is first an idea and later a passion. Rather, the initiating idea and the sympathetic passion are numerically distinct. Hume may mean to express this view, when he writes that “the ideas of the affections are converted into the very impressions they represent, and … the passions arise in conformity to the images we form of them” (T 2.1.11.8/319, emphasis added). It seems odd to say that a single perception “arise[s] in conformity to” itself. So, this passage suggests that the initiating idea (or “image”) and the sympathetic passion are two perceptions, not a single perception that has changed.
If this second interpretation is correct, the initiating idea is numerically distinct from the sympathetic passion. So, the fact that the former is a copy cannot be used to show that the latter is a copy as in the argument above. However, on this second interpretation, the sympathetic passion is caused by and closely resembles the initiating idea. It is caused by the initiating idea: if one had not first formed an initiating idea that represents a certain type of passion, one would not have formed a sympathetic passion of that type. And it closely resembles the initiating idea: the initiating idea is copied from an earlier passion, which it exactly resembles, save for its lower degree of liveliness and its representationality; the sympathetic passion exactly resembles the initiating idea, save for its higher degree of liveliness and its lack of representationality; so, the sympathetic passion resembles the initiating idea just as closely as does the earlier passion from which the initiating idea is copied. So, the sympathetic passion is a copy of the initiating idea. And so, if our second interpretation of “conversion” is correct— that is, if the initiating idea is succeeded by the sympathetic passion—then, again, Hume must (for consistency) accept that sympathetic passions are copies.29
Objection:30 Unlike the argument given in connection with our first interpretation of “conversion,” this argument does presuppose a specific view of what copying is: namely, that copying consists in causality and close resemblance. This view is often attributed to Hume,31 and may be his official view,32 but— the objection continues—it is too crude for his own purposes. Certain kinds of causal connection between closely resembling things are deviant, hence do not suffice for copying. Sympathetic passions closely resemble and are caused by initiating ideas, but the causal chain in this case is deviant. So, Hume need not accept that sympathetic passions are copies.
Reply:33 I doubt that Hume could consistently spell out the deviant/non-deviant contrast in a way that meets his needs—or, better, the needs he would have, if he accepted CTR. No feature of the causal chain involved in sympathy (hereinafter, the sympathy chain) clearly marks it as deviant. Two features differentiate this chain from a paradigm case of Humean copying where a simple impression causes a resembling simple idea. First, the paradigm case involves loss of liveliness (the simple idea is less lively than the simple impression), whereas the sympathy chain involves gain (the sympathetic passion is livelier than the initiating idea). However, it is not clear that this difference gives Hume any grounds for classifying the sympathy chain as deviant: if a less lively perception can count as a copy of a livelier one (as in the paradigm case), why can’t a livelier perception count as a copy of a less lively one (as in sympathy)?
Second, the paradigm case involves only two perceptions (the simple impression and the simple idea), whereas the sympathy chain involves three: the initiating idea, the perception of self, and the sympathetic passion. But if Hume accepts CTR, then he cannot consistently classify the sympathy chain as deviant just in virtue of its having this three-perception structure. On the interpretation we are now considering, copying is defined in terms of close resemblance and a non-deviant causal chain. So, if Hume classifies the sympathy chain as deviant in virtue of its having this three-perception structure, he thereby classifies a sympathetic passion as not a copy. However, if Hume accepts CTR, then he cannot consistently say that a perception is not a copy just in virtue of its resulting from a causal chain with this three-perception structure. To see this, recall that probable reasoning shares the three-perception structure of the sympathy chain: a belief formed via such reasoning is copied from one past perception, but receives its high degree of liveliness from a third perception. (In our schematic example above, Smith’s belief that a B will occur is copied from a past impression of a B but enlivened by a present impression of an A.) If Hume accepts CTR, then he cannot say that beliefs resulting from probable reasoning are not copies, on pain of inconsistency with his views that beliefs are ideas (T 1.3.7.5/96) and that all ideas are representational (T 1.2.3.11/37, 1.3.14.6/157). So, if he accepts CTR, then he cannot consistently say that a perception is not a copy just in virtue of its resulting from a causal chain with the three-perception structure common to probable reasoning and the sympathy chain. And so, he cannot say that the sympathy chain is deviant just in virtue of its having this structure.
So, neither of the sympathy chain’s distinctive features provides Hume with grounds on which he can consistently classify it as deviant. And so, it is not clear that he can draw a deviant/non-deviant distinction that would help him avoid the conclusion that sympathetic passions are copies.
I conclude that, whichever interpretation of sympathetic “conversion” is correct—that is, whether initiating ideas become or are succeeded by sympathetic passions—if Hume accepts CTR, then he must accept that sympathetic passions are copies. But he must also accept that they are non-representational because he denies that any passion is representational. When taken together with his view that passions are simple perceptions, CTR implies that a passion is representational if it is a copy. So, Hume cannot consistently accept CTR, given Treatise Book 2’s account of the passions.
My argument in Section 3 rests on the claim that, for Hume, passions are not representational. This claim seems to be well supported by the Representation Argument, where we have seen Hume write that a passion “contains not any representative quality” and that feeling a passion involves no “reference to any other object” (T 2.3.3.5/415). However, some scholars think that Hume has good reason, independently of my argument, to strike out this passage. For example, Annette Baier (1991: 160–62) famously calls the Representation Argument a “very silly paragraph” that does not cohere with the rest of Hume’s theory of the passions. Those wishing to resist my argument in Section 3 might therefore say:34 since we already knew Hume has good independent reason to strike out the Representation Argument, it is not troubling to find that CTR conflicts with it (when it is taken together with other elements of Hume’s theory of the passions); this result just gives Hume all the more reason to jettison the Representation Argument; having done so, he can consistently accept CTR.
I have three replies to this manoeuvre. First, it undermines the evidence that Hume accepts CTR. For the Representation Argument is an important piece of that evidence. More specifically, it is an important piece of the evidence that Hume accepts CTR’s answer to Garrett’s (2006: 302) question “why … perceptions represent—that is, what makes them representational at all.” CTR says: a perception is representational if and only if it is a copy. The Representation Argument is the main piece of evidence that Hume accepts this conditional’s left-to-right direction.
Moreover, when Rachel Cohon and David Owen (1997) proposed that Hume accepts CTR, in their seminal paper on his theory of mental representation, their main goal was to interpret the Representation Argument. The appeal of attributing CTR to Hume was, precisely, to make sense of his otherwise surprising claim, in the course of that argument, that passions are non-representational.
So, even if there are good independent reasons for Hume to strike out the Representation Argument, those who think he accepts CTR cannot satisfactorily defend their interpretation by appealing to this fact.
Second, there are not, in fact, good independent reasons for Hume to strike out the Representation Argument. Baier (1991: 160) called this argument a “very silly paragraph” because she thought she saw a conflict between (i) its claim that passions involve no “reference to any other object” and (ii) Hume’s view that passions have “objects” on which they are “directed” (T 2.1.3.4/280, 2.2.1.2/329)— in other words, that passions exhibit intentionality. However, as others have observed,35 there is no genuine conflict here. For Hume, to say that a passion has an object is not to say that the passion itself represents that object. Rather, it is to say that the passion is suitably related to an idea that represents that object.36 On one version of this view, “suitably related” is cashed out in causal terms, as follows. When Hume attributes an “object” to a direct passion such as desire or fear (T 2.3.3.3/414, 2.3.9.27/446), he means that the passion is caused by an idea that represents that object.37 And when he attributes an “object” to an indirect passion such as pride, he means that the passion is embedded in a complicated causal chain (the “double association” of impressions and ideas) culminating in an idea of that object. For example, when he says that the “object” of pride is the self, he means that pride, having arisen from a cause related to oneself and a separate pleasure deriving from that cause, then produces or “turns our view to” an idea that represents oneself (T 2.1.2.4/278).38
So, the objection to my argument has a false premise: Hume’s view that passions have “objects” is not a good reason, independent of my argument in Section 3, for him to strike out the Representation Argument.
Third and more concessively, I am sympathetic to the claim that Hume should abandon the Representation Argument and I will argue that he does abandon it in his mature philosophical works (Section 6). But his doing so is part of a package of changes leaving no clear evidence that he accepts CTR. So, it provides no comfort to those who attribute CTR to him.
So far, I have argued that the Hume of the Treatise cannot consistently accept CTR. Karl Schafer (2015) and Donald Ainslie (2015; 2019) deny that Hume accepts CTR, but nonetheless see important roles for copying in his theory of mental representation. I will now argue that their interpretations face similar problems to CTR: neither can be squared with Hume’s account of sympathetic passions.
Schafer (2015: 990–98) argues that Hume accepts a “hybrid” theory of mental representation that appeals to both copying and “functional role”: copying explains an “imagistic” kind of representation, that is, the representation of an object and its intrinsic qualities by means of resemblances between a perception and what it represents; functional role explains how a perception that “imagistically” represents an object and its intrinsic qualities can also non-imagistically represent extrinsic “formal, relational or structural” features of objects. For example, a particular idea might imagistically represent a particular object as having a certain specific triangular shape (an intrinsic quality, let us suppose) in virtue of being copied from a triangle-impression. An abstract idea, on the other hand, might both represent a particular object as having this specific shape, while also representing “all the particular triangles that fall within [the category of triangles] as resembling each other in certain respects that go beyond the sort of resemblance that is a straight-forward product of their intrinsic qualities” (Schafer 2015: 992). Insofar as this abstract idea’s representational content involves more than what is represented “imagistically,” it is explained by the idea’s functional role—specifically, by the idea’s associations with general terms and other ideas (Schafer 2015: 992–93).
Schafer does not explicitly distinguish Garrett’s (2006: 302) questions, “what makes [perceptions] representational at all” and “what determines [perceptions’] specific representational content,” as we have done. However, I understand the hybrid theory to address both questions: copying explains both what makes perceptions representational at all and some aspects of a representational perception’s specific representational content (namely, the “imagistic” aspects); functional role explains the other (non-imagistic) aspects of a representational perception’s specific representational content.39
So understood, the hybrid theory faces much the same problem as CTR. It entails that if a perception is a copy, then it is an imagistic representation. But sympathetic passions are not representations of any kind,40 despite their being copies.41 So, Hume cannot accept the hybrid theory consistently with Book 2’s account of sympathetic passions.
CTR and the hybrid theory are both naturalistic in the sense explained above (Section 1): both aim to “account for the intentionality of mental states” without positing any “explanatorily basic intentional properties” (Garrett 2006: 302). In contrast, on Donald Ainslie’s (2015: 211–17; 2019) reading, Hume accepts a theory according to which there is an explanatorily basic kind of intentionality. This theory posits two kinds of copying. To explain their roles, I must first briefly explain Ainslie’s views about the nature of Humean impressions.43
Ainslie (2015: 212) claims that sensory impressions44 have “two inseparable aspects”: an image-content and an act of awareness.45 A sensory impression’s image-content is what we are aware of, in having that impression. Ainslie claims that, “most primitively,” image-contents are “unextended coloured or tangible points, or non-spatial smells, tastes, and the like” (2015: 211–12). (Ainslie’s Hume does not explain impressions’ having these “most primitive” forms of image-content in terms of their non-intentional properties and relations, hence his theory is not naturalistic in our sense.)46 The other aspect of a sensory impression is the “action of the mind” by which we are aware of the image-content (2015: 212). These two aspects of a sensory impression are “inseparable” (2015: 212): an image-content’s esse is percipi (2015: 212n29), so there cannot be an image-content without an act of awareness of it; conversely, there cannot be an act of sensory (n.b.) awareness “without it being the awareness of image-content” (2015: 212).47 For Ainslie (2015: 213), a perception’s degree of liveliness is a feature of its act-aspect, not of its image-content.
Ainslie thinks that passions differ significantly from sensory impressions. As we have seen, Hume claims that a passion “contains not any representative quality” (T 2.3.3.5/415). Ainslie infers that, unlike sensory impressions, passions have no image-contents: they are “nothing but different ‘flavours’ of vivacity” (2015: 213), which do not present anything to the mind.
Ainslie’s Hume posits two kinds of copying in order to explain how different kinds of ideas result from impressions. The first kind of copying produces a “primary idea” that exactly resembles the impression, save for its degree of liveliness. When applied to a sensory impression with an image-content, this kind of copying produces an idea with the same specific type of image-content (2015: 123). For example, when applied to a visual impression whose image-content is three red points arranged triangularly, this kind of copying produces an idea that also has, as its image-content, three red points arranged triangularly. When applied to a passion with no image-content, this kind of copying produces an idea with no image-content: a “non-representational idea” that is merely “a less vivacious version of the mental qualification that is the original passion” (Ainslie 2019).
The second kind of copying produces a “secondary idea” whose image-content is the “primary perception,” that is, the impression or primary idea, from which it derives (Ainslie 2015: 123). For example, a philosopher who introspectively examines one of her perceptions does so by forming a “secondary idea” whose image-content is the perception from which it derives. When a secondary idea derives from a sensory impression or a primary idea copied from a sensory impression, the secondary idea’s image-content includes both the image-content and the act-aspect of that other perception. When a secondary idea derives from a passion, it is an idea “of” that passion: in other words, its image-content is “the [non-representational] mental qualification that is the original passion” (Ainslie 2019).
Hume’s account of sympathy presents a series of nested choice-points for Ainslie. By working through them, I will argue that they reveal Ainslie’s interpretation to be untenable.
First, is the initiating idea (1) a primary idea derived from a previous passion via the first kind of copying, or (2) a secondary idea derived via the second? Ainslie (2019) favours option (1). When applied to a passion with no image-content, the first kind of copying produces a primary idea with no image-content: a “non-representational idea” (Ainslie 2019). It follows that the initiating idea is “non-representational.” But this result conflicts with Hume’s claims both about ideas generally and about initiating ideas specifically. Hume claims that all ideas “represent”: the Copy Principle incorporates the claim that every simple idea “exactly represent[s]” its correspondent simple impression (T 1.1.1.7/4); and elsewhere Hume extends this claim to all ideas, writing for example that “Ideas always represent the objects or impressions, from which they are deriv’d” (T 1.2.3.11/37), and that “Ideas always represent their objects or impressions” (T 1.3.14.6/157). And in the Treatise section “Of the love of fame,” Hume says explicitly that the initiating ideas involved in sympathy are “images” that “represent” or are “of” passions: he writes that sympathy begins with “an idea of” the “affection” that is “infus’d” into the sympathizer (T 2.1.11.3/317); that “the ideas of the affections of others are converted into the very impressions they represent” (T 2.1.11.8/320); and that initiating ideas are “images … of” other people’s passions (T 2.1.11.8/319).
Ainslie should therefore prefer option (2), on which the initiating idea is a secondary idea derived from a previous impression via the second kind of copying—that is, an idea whose image-content is the perception from which it derives. He now faces a further choice-point, based on the two ways to construe Hume’s account of sympathetic “conversion” (Section 3):48 is the initiating idea (2A) a secondary idea that becomes a sympathetic passion, or (2B) a secondary idea that is succeeded by a sympathetic passion? Option (2A) gives rise to a mystery. For Ainslie, a secondary idea has an image-content, but a sympathetic passion does not. So (2A) commits him to saying that, in the course of sympathetic communication, a perception that has an image-content (when it is an initiating idea) ceases to have an image-content (when it becomes a sympathetic passion) simply by gaining liveliness. But how can this be? For Ainslie, a perception’s degree of liveliness is a feature of its act-aspect, not of its image-content. How can modifying one aspect of a perception (its act-aspect/liveliness) destroy its other aspect (its image-content)? This upshot is mysterious.
Ainslie should therefore prefer option (2B), on which the initiating idea is succeeded by the sympathetic passion. He now faces the third and last choice-point: Is the sympathetic passion (2Bi) a copy of the initiating idea, or (2Bii) not a copy of it? If Ainslie opts for (2Bi), then he must accept that the sympathetic passion has an image-content. We are currently supposing, on Ainslie’s behalf, that (2) the initiating idea is a secondary idea, hence has an image-content. When applied to a perception with an image-content, both kinds of copying generate a further perception with an image-content. So, if (2Bi) the sympathetic passion is a copy (either kind of copy) of the initiating idea, then it too has an image-content. But this result conflicts with Ainslie’s interpretation of the Representation Argument, which says that no passion has an image-content.
Suppose, then, that Ainslie opts for (2Bii), on which the sympathetic passion is not a copy (that is, not either kind of copy) of the initiating idea. We are currently supposing, on Ainslie’s behalf, that (2B) the initiating idea is succeeded by the sympathetic passion. I have argued that, on this interpretation of sympathetic “conversion,” the sympathetic passion is caused by and closely resembles the initiating idea (Section 3). So, if the sympathetic passion is (2Bii) not a copy of the initiating idea, then causation and close resemblance are not sufficient for copying (that is, for either kind of copying that Ainslie distinguishes). Ainslie now owes us an answer to the following question. What is sufficient for copying: what other relation(s) must be added to causation and close resemblance, in order for there to be a case of one or the other kind of copying? Hume posits only seven basic types of philosophical relations (T 1.1.5) and, setting aside causation and resemblance, none of the remaining five seems relevant to what copying is: a copy is not identical to its original (T 1.1.5.4/14); there is no specific spatial or temporal relation in which a copy must stand to its original (T 1.1.5.5/14); there is no particular quantity of copies that must be produced from an original, in order for any one of them to count as a copy (T 1.1.5.6/14–15); there need be no difference in degree of any quality between a copy and its original (T 1.1.5.7/15); and it is unclear how the remaining relation, contrariety, could be used to explain what copying is (T 1.1.5.8/15). So, if Ainslie wishes to say that (2Bii) the sympathetic passion succeeds but is not a copy of the initiating idea, he must say that the two kinds of copying he distinguishes are sui generis relations that cannot be analysed in terms of causation, resemblance, or other of the philosophical relations in Hume’s typology. Whatever this move’s philosophical merits, it is unattractive as an interpretation of the Treatise. If Hume meant to posit two sui generis kinds of copying and accord them important roles in his theory of mental representation, he would presumably have included them in his typology of relations (T 1.1.5)—but he does not.
To summarize, Hume’s account of sympathy presents only unsatisfactory options for Ainslie: either (1) the initiating idea is a non-representational primary idea with no image-content, which conflicts with the texts; or (2A) the initiating idea is a representational secondary idea that becomes a non-representational sympathetic passion simply by gaining liveliness, which is mysterious; or (2Bi) the initiating idea is a representational secondary idea and the sympathetic passion is a copy of it, which implies that the sympathetic passion is representational, contrary to Ainslie’s view of Humean passions; or, lastly, (2Bii) the initiating idea is a representational secondary idea and the sympathetic passion is not a copy of it, which leaves Ainslie with no satisfactory account of what either kind of copying is. I conclude that, like the others we have considered, the theory of mental representation that Ainslie finds in the Treatise cannot be squared with Hume’s account of sympathetic passions.
Scholars who think Hume accepts CTR may accept the arguments above and conclude that Hume should jettison Book 2’s account of sympathy. To support this conclusion, they may point to Hume’s later philosophical works, which contain a significantly different view of sympathy. I will argue that, on the contrary, Hume’s later works present a further challenge to their interpretation: Hume does revise his views about sympathy; but these revisions are part of a package of changes that leaves no clear evidence that he accepts CTR. This challenge is addressed to scholars who hold that Hume himself actually accepts CTR49—a group that seems to include Cohon and Owen (1997), Weintraub (2005), and Landy (2012; 2015).50
Hume’s treatment of sympathy in the moral Enquiry differs strikingly from that in the Treatise. As we have seen, the Treatise tries to explain sympathy in terms of Hume’s principles of association and the transfer of liveliness among associated perceptions. In contrast, the moral Enquiry gives no account of the mechanism of sympathy and indeed suggests we cannot give one. Hume notes that “there are, in every science, some general principles, beyond which we cannot hope to find any principle more general,” and suggests that our tendencies to be sympathetically pleased by others’ happiness and pained by their misery are examples: “It is not probable, that these principles can be resolved into principles more simple and universal … we may here safely consider these principles as original” (M 5.17n/219–20n).
Moreover, Hume weakens his claims about the products of sympathy. Another person’s happiness, he now says, leads us via sympathy to feel “pleasure,” another’s misery to feel “pain” (M 5.17n/219–20n). He says nothing to suggest that our sympathetic pleasure (pain) must be an impression of the same specific type as the happiness (misery) with which we sympathise.
Because of these changes, the Hume of the moral Enquiry is not committed to saying that sympathy produces copies of previous perceptions. So, he is not committed to there being non-representational copies. And so, his view of sympathetic passions no longer commits him to rejecting CTR. Hume’s revised account of sympathy might therefore seem to play into my opponents’ hands.
However, other differences between Hume’s later works and the Treatise undermine the view that the mature Hume accepts CTR. The main pieces of evidence that the Hume of the Treatise accepts CTR are, first, the various passages that introduce, defend, and apply the Copy Principle; and second, the Representation Argument (Section 2). But the later works contain no counterparts of these pieces of evidence: the first Enquiry’s Copy Principle differs from that of the Treatise, and does not provide evidence that Hume accepts CTR; and the Representation Argument is entirely missing from Hume’s discussions of motivation and moral evaluation in the Dissertation and the moral Enquiry. Let us consider each point in turn.
As we have seen (Section 2), the Treatise’s Copy Principle says that every simple idea is caused by (“deriv’d from”), resembles (is “correspondent to”), and is a representation of (“exactly represents”) a simple impression (T 1.1.1.7/4). Hume explicitly argues only that every simple idea is caused by and resembles a simple impression (T 1.1.1.5–6/3–4, 1.1.1.8–9/4–5), and claims to have thereby “establish[ed]” the Copy Principle (T 1.1.1.7/4, 1.1.1.12/7). So, his formulation of the Copy Principle and his way of introducing and motivating it suggest that he considers simple ideas “represent[ations]” of impressions, and that he regards a simple idea’s being copied from (that is, resembling and being caused by) a simple impression as sufficient for its representing that impression, in keeping with CTR.
In contrast, consider how Hume introduces the first Enquiry’s Copy Principle:
In short, all the materials of thinking are derived either from our outward or inward sentiment: The mixture and composition of these belongs alone to the mind and will. Or, to express myself in philosophical language, all our ideas or more feeble perceptions are copies of our impressions or more lively ones. (E 2.5/19)51
This passage does not say that ideas “represent” impressions.52 Moreover, it suggests that Hume’s talk of ideas’ being “copies” of impressions is dispensable—a highbrow (“philosophical”) way of saying something equally expressible in more down-to-earth language, that is, that “all the materials of thinking are derived either from our outward or inward sentiment.” However, this claim is ambiguous. By ‘the materials of thinking’, Hume might mean the elements of the objects of thought, that is, what we think about. So interpreted, the “materials” of the thought of a virtuous horse are virtue and a horse. Alternatively, he might mean the elements of the mental representations that (supposedly) enable us to think about those things. So interpreted, the “materials” of the thought of a virtuous horse are a mental representation of virtue and a mental representation of a horse.
When Hume reintroduces the Copy Principle in Enquiry §7, he disambiguates the earlier passage for us:
It seems a proposition, which will not admit of much dispute, that all our ideas are nothing but copies of our impressions; or, in other words, that it is impossible for us to think of any thing, which we have not antecedently felt, either by our external or internal senses. (E 7.4/62, boldface added)
By his use of the phrase ‘or in other words’,53 Hume suggests that the Enquiry’s Copy Principle—“all our ideas are nothing but copies of our impressions”— is another way of expressing a claim about the objects of thought and feeling: “it is impossible for us to think of any thing, which we have not antecedently felt, either by our external or internal senses.” Hume’s claim seems to be that, as a matter of psychological law, we can think of x only if we have previously “felt” x (that is, had first-hand experience of x). By itself, this claim is silent about the mental representations that (according to CTR) enable us to think of things by representing them. Consequently, by itself, the Enquiry’s Copy Principle is silent about both “what makes [a mental representation] representational at all” and about “what determines [a mental representation’s] specific representational content.” So, this principle does not constitute a theory of mental representation, in the sense explained above (Section 1). And so, it does not constitute evidence that Hume accepts CTR, which is a theory of mental representation in that sense.
My opponents may reply that by accepting the Enquiry’s Copy Principle, as I have interpreted it, Hume commits himself to accepting CTR as well. To support this claim, they may argue as follows. The Enquiry’s Copy Principle places a constraint on the possible objects of thought: namely, that we cannot “think of any thing, which we have not antecedently felt” (E 7.4/62). This constraint demands an explanation. Unless Hume can provide one, he is unjustified in accepting the Enquiry’s Copy Principle. Accepting CTR would allow him to explain the constraint. Absent any other explanation of equal or greater plausibility, he is therefore justified in accepting the Enquiry’s Copy Principle only if he also accepts CTR. So, for reasons of interpretive charity, we should suppose that he actually does accept CTR.54
However, this argument is unpersuasive because Hume would reject one of its premises: namely, the premise that he is unjustified in accepting the Enquiry’s Copy Principle unless he can explain the constraint it places on the possible objects of thought. Hume supports the Enquiry’s Copy Principle via two empirical arguments based on everyday observational evidence that we cannot think of anything we have not previously felt (E 2.6–7/19–20). He says nothing to explain why this constrain obtains. As in the case of sympathy (M 5.17n/219–20n), he now seems content to accept this constraint as “original,” without trying to “resolve[] it into principles more simple and universal”—for example, into the principles about mental representation that comprise CTR. Nevertheless, he holds that his two empirical arguments provide adequate justification for accepting the Enquiry’s Copy Principle: he writes that they are “sufficient” to “prove” this principle in the absence of any counterexamples that would require him to revise it (E 2.6/19, 2.8/20).55
I conclude that, in contrast with the Treatise’s Copy Principle, the first Enquiry’s Copy Principle does not provide evidence that Hume accepts CTR.
Let us now consider the other main piece of evidence that the Hume of the Treatise accepts CTR: the Representation Argument. In the Treatise, Hume seems to hold this argument in high esteem: Books 2 and 3 both contain versions of it (T 2.3.3.5/415, 3.1.1.9/458), and Hume rests his main claims about motivation and moral evaluation on it. But it does not appear in the “recast” versions of these books: the Dissertation on the Passions and the moral Enquiry.
Section 5 of the Dissertation, which roughly corresponds to the Treatise section “Of the influencing motives of the will” (T 2.3.3), gives only a compressed version of Hume’s other main argument in that section: neither relations of ideas nor matters of fact can motivate; so, reason “of itself” cannot motivate (P 5.1; cf. T 2.3.3.2–3/413–14). This argument says nothing about the representationality or non-representationality of the passions. So, Hume’s use of it provides no evidence that he accepts CTR. Section 1 of the moral Enquiry gives a similar version of the same argument (M 1.7/172), and Appendix 1 gives a new argument about reason and motivation, concerning a regress of reasons for action (M App 1.18–20/293–94), which again says nothing about the representationality or non-representationality of the passions. Neither work contains the Representation Argument or any argument recognisably descended from it.
I conclude that, in contrast with the Treatise, Hume’s later philosophical works provide no clear evidence that he accepts CTR. Hume famously urged his readers to regard these works “alone … as containing his philosophical sentiments and principles” (E Advertisement/xlii). So, there is no clear evidence that CTR is among his considered philosophical views.
Recent decades have seen much debate, focused on the Treatise, about Hume’s theory of mental representation. The most popular proposal is that Hume accepts CTR (Section 2) or another theory in which copying plays a significant explanatory role (Section 5). I have presented two challenges to these proposals: none of the interpretations considered here can be squared with Book 2’s account of sympathetic passions (Sections 3–5); and Hume’s later philosophical works provide no clear evidence that he accepts CTR (Section 6).
Do Hume’s works contain another theory of mental representation, according to which something other than copying explains “what makes [perceptions] representational at all” and “what determines [a representational perception’s] specific representational content”? For example, do they—as Don Garrett (2006; 2015) has proposed—contain a theory that explains mental representation in terms of causal and functional relations? I must leave these questions for another occasion.
I presented a version of Sections 2 and 3 at the 44th Hume Society Conference held at Brown University in July 2017, and versions of the whole paper at the University of Edinburgh and at the Aspects of British Early Modern Philosophy Conference held at St Peter’s College, Oxford, in September 2019. My comments on Donald Ainslie’s Hume’s True Scepticism, at the 2017 Pacific APA meeting, included a predecessor of Section 5. Many thanks to the audiences on all those occasions; to my fellow symposiasts at the APA; and especially to David Landy (my commentator at the Hume Conference) and Donald Ainslie for much helpful feedback and advice—and, in Ainslie’s case, for sharing and allowing me to cite unpublished work related to the topic of this paper. Thanks also to Matthew Chrisman, Rachel Cohon, Jade Fletcher, Don Garrett, Alistair Isaac, David Levy, Peter Millican, Bryan Pickel, Lewis Powell, Hsueh Qu, Alasdair Richmond, Mike Ridge, Mark Sprevak, and several anonymous referees.
Ainslie, Donald C. (2005). Sympathy and the Unity of Hume’s Idea of Self. In Joyce Jenkins, Jennifer Whiting, and Christopher Williams (Eds.), Persons and Passions: Essays in Honor of Annette Baier (143–73). University of Notre Dame Press.
Ainslie, Donald C. (2015). Hume’s True Scepticism. Oxford University Press.
Ainslie, Donald C. (2019). Comments on Elizabeth Radcliffe’s Hume, Passion, and Action. Comments presented at the 46th International Hume Society Conference held at the University of Nevada–Reno, July 22–26, 2019.
Baier, Annette C. (1991). A Progress of Sentiments: Reflections on Hume’s Treatise. Harvard University Press.
Cohon, Rachel and David Owen (1997). Hume on Representation, Reason, and Motivation. Manuscrito, 20(2), 47–76.
Cottrell, Jonathan (2018). Hume on Mental Representation and Intentionality. Philosophy Compass, 13(7), e12505. https://doi.org/10.1111/phc3.12505https://doi.org/10.1111/phc3.12505
Cottrell, Jonathan (in press). Perceptions, Minds, and Hume’s Self-Doubts. Hume Studies.
Cummins, Robert (1989). Meaning and Mental Representation. MIT Press.
Everson, Stephen (1988). The Difference between Feeling and Thinking. Mind, 97(387), 401–13.
Fodor, Jerry A. (2003). Hume Variations. Oxford University Press.
Garrett, Don (1997). Cognition and Commitment in Hume’s Philosophy. Oxford University Press.
Garrett, Don (2006). Hume’s Naturalistic Theory of Representation. Synthese, 152, 301–19.
Garrett, Don (2015). Hume. Routledge.
Hamid, Nabeel. (2015). Hume’s (Berkeleyan) Language of Representation. Hume Studies, 41(2), 171–200.
Hume, David (1975). Enquiries concerning Human Understanding and concerning the Principles of Morals (3rd ed., L. A. Selby-Bigge and P. H. Nidditch, Eds.). Clarendon Press. (Original work published 1748–51)
Hume, David (1978). A Treatise of Human Nature (2nd ed., L. A. Selby-Bigge and P. H. Nidditch, Eds.). Clarendon Press. (Original work published 1739–40)
Hume, David (1998). An Enquiry concerning the Principles of Morals (Tom L. Beauchamp, Ed.). Clarendon Press. (Original work published 1751)
Hume, David (2000). An Enquiry concerning Human Understanding (Tom L. Beauchamp, Ed.). Clarendon Press. (Original work published 1748)
Hume, David (2007a). A Treatise of Human Nature: Vol. 1. Texts (David Fate Norton and Mary J. Norton, Eds.). Clarendon Press. (Original work published 1739–40)
Hume, David (2007b). A Dissertation on the Passions; The Natural History of Religion (Tom L. Beauchamp, Ed.). Clarendon Press. (Original work published 1757)
Landy, David (2012). Hume’s Theory of Mental Representation. Hume Studies, 38(1), 23–54.
Landy, David (2015). Kant’s Inferentialism: The Case Against Hume. Routledge.
Landy, David (2018). Recent Scholarship on Hume’s Theory of Mental Representation. European Journal of Philosophy, 26(1), 333–47. http://dx.doi.org/10.1111/ejop.12245http://dx.doi.org/10.1111/ejop.12245
Merivale, Amyas (2009). Hume’s Mature Account of the Indirect Passions. Hume Studies, 35(1, 2), 185–210.
Qu, Hsueh (2012). The Simple Duality: Humean Passions. Canadian Journal of Philosophy, 42(S1), 98–116.
Ramsey, William M. (2007). Representation Reconsidered. Cambridge University Press.
Schafer, Karl (2015). Hume’s Unified Theory of Mental Representation. European Journal of Philosophy, 23(4), 978–1005.
Stroud, Barry (1977). Hume. Routledge and Kegan Paul.
Weintraub, Ruth (2005). A Humean Conundrum. Hume Studies, 31(2), 211–24.
Weller, Cass (2002). The Myth of Original Existence. Hume Studies, 28(2), 195–230.