In his famous paper “What Is It Like To Be a Bat?” Thomas Nagel argues that we are severely limited when it comes to understanding subjective perspectives that are radically different from our own. According to Nagel, the only way someone can presently think about the phenomenal character of experiences that they haven’t had comes via imagination. Such imaginative efforts fall short when the phenomenal character is radically different from anything the individual has previously experienced, as is the case when a human tries to understand what it’s like to be a bat.

Of course, when it comes to trying to imagine what it’s like to be a bat, we’re hindered by the fact that we have no way of asking the bat what its experiences are like. Humans can’t have that sort of meaningful communication with bats. But even if we could, Nagel doesn’t seem to think that would be enough to enable our imagination to do what it would need to do. The same kind of problem arises even in cases where communication is possible, as when someone blind from birth talks to a sighted person in an effort to help them imagine what it’s like to see.Footnote 1 As Nagel notes, “The loose intermodal analogies – for example, ‘Red is like the sound of a trumpet’ – which crop up in discussions of this subject are of little use.” (1974, 449).

As he goes on to argue, however, it might be possible in principle to overcome this problem. To enable us to understand what an experience is like when we are not able to imaginatively take the point of view of the experiential subject, Nagel suggests that we should work to develop an objective vocabulary for describing phenomenology:

This should be regarded as a challenge to form new concepts and devise a new method – an objective phenomenology not dependent on empathy or the imagination. Though presumably it would not capture everything, its goal would be to describe, at least in part, the subjective character of experiences in a form comprehensible to beings incapable of having those experiences. (1974, 449)

It’s not clear what such vocabulary would look like. Perhaps it’s what Paul and Patricia Churchland have in mind when they speculate about a future way of describing mental states in which “whole chunks of English… are replaced by scientific words that call a thing by its proper name” (MacFarquhar 2007, 68). In fact, the Churchlands already try to conceptualize their experiences in this objective manner; for example, when Paul cuts himself shaving, he distinguishes the A-delta-fiber pain from the C-fiber pain, and Pat describes her feelings of frustration in terms of serotonin levels and a flood of glucocorticoids (MacFarquhar 2007, 68–69). But whatever form such an objective phenomenological vocabulary would take, Nagel is optimistic about the prospects, suggesting that “it should be possible to devise a method of expressing in objective terms much more than we can at present, and with much greater precision.” (1974, 449).

Nagel’s discussion thus proposes that the problem arises only when we are trying to communicate in third-person terms. It arises from our lack of an objective vocabulary for talking about experience – a vocabulary available to someone who does not occupy the relevant point of view and thus does not have the relevant experiences themself. But I think the problem runs deeper than that. To my mind, we also lack an adequate subjective vocabulary for describing phenomenology. Even when we do occupy the relevant point of view, and we ourselves have the relevant experiences, the phenomenological vocabulary available to us to describe them falls short.

Though Nagel suggests that subjective concepts afford us an “ease of description” (1974, 449), I think that we often struggle to describe our own phenomenal states in terms we ourselves find adequately expressive. Even when we are communicating with people whom we can reasonably expect to have had similar experiences to our own, we often find ourselves surprisingly unable to provide a meaningful characterization of what the experiences are like – a phenomenon that’s perhaps especially salient when people are in their doctors’ offices and asked to describe their present experiences of pain. Where exactly is it? Is it sharp or dull? Heavy? Radiating? Throbbing? Though sometimes words like these might help to convey part of the story, they too often seem to fall woefully short of communicating what the pain is like. Even worse, these descriptors often seem patently inapplicable, and more fitting words aren’t there. Our language just doesn’t seem up to the task.Footnote 2

To my mind, the focus on bats has tended to obscure the breadth of the problem. We tend to think of the problem of describing our phenomenology as one that arises when we’re looking for words that enable us to communicate across vast experiential divides. But in fact, our phenomenological vocabulary is strikingly impoverished even when it comes to communicating with others who are on the same side of the divide – or even when it comes to capturing things for ourselves. This limitation of our vocabulary is what I call the impoverishment problem. In this paper, I aim to explore this problem and the possibilities for a solution to it.Footnote 3

The discussion begins in Part I with an attempt to clarify the problem by comparing the notion of impoverishment to the notion of ineffability that often arises in discussions of qualia. Having done so, I turn in Part II to look at some specific examples drawn from both philosophy and ordinary life where the impoverishment of our phenomenological vocabulary is on display. I focus on two examples in particular: first, phenomenological descriptions of remembering, and second, phenomenological descriptions of pain. With these examples before us, in Part III I show why the problem cannot be easily explained away or reduced to other familiar problems. As the discussion of this section suggests, the impoverishment problem is both real and robust. That said, the discussion also shows that we need not take the problem to be in principle unresolvable. Thus, in Part IV, I offer some suggestions for how we might go about making progress on solving it. I end with some brief concluding remarks about some wider-reaching philosophical ramifications of the matters discussed in this paper.

1 Clarifying the problem

In a famous discussion of qualia, Daniel Dennett has suggested that the standard philosophical treatment of qualia involves four claims: Qualia are (1) ineffable; (2) intrinsic; (3) private; and (4) directly or immediately apprehensible to consciousness (Dennett, 1988). But if this is right, then it might seem that the answer to the impoverishment problem is obvious – or, better put, that it’s obvious that there can be no answer to the impoverishment problem. Our phenomenological vocabulary is impoverished because qualia, by definition, cannot be described. Capturing them in language is in principle impossible. The impoverishment of our phenomenological vocabulary should not be seen as something that is surprising or problematic, nor should it be seen as something that can be overcome.

Unsurprisingly, perhaps, I don’t think matters are quite so simple or obvious. First, as many philosophers have pointed out, we can talk about qualia in the sense of the “what it’s likeness” of our experience without buying into the claim that they have the four features involved in Dennett’s characterization (see, e.g., (Levin, 2011), (Tye, 2021). Whether qualia must have the metaphysical and epistemological characteristics built into Dennett’s characterization is actually a substantive matter that must be argued. Second, even if were to turn out that qualia are strictly speaking ineffable, that would not entirely explain the impoverishment of our phenomenological vocabulary. We might still be able to provide significant and substantive characterizations of our qualitative states, even if the words were eventually to run out before we reached a fully complete description.

As evidence that philosophers don’t treat phenomenal experience as fundamentally indescribable, we can point to a large number of different philosophical contexts in which philosophers take up the task of describing such experience. Particularly in philosophy of mind, there are numerous efforts at descriptive phenomenology, i.e., attempts to characterize the phenomenal character of various kinds of experiences. Philosophers working on imagination, perception, and memory have offered various treatments of the phenomenology of these mental activities, as well as explanations of how they can be phenomenally differentiated from one another. Discussions comparing and contrasting the different sense modalities often take up the question of the phenomenal differences among them and, more specifically, whether these phenomenal differences can be adequately accounted for in representational terms. Philosophers arguing for the existence of cognitive phenomenology attempt to describe its phenomenal character and argue that it cannot be given a reductive treatment. As these examples suggest, philosophical discussion of our phenomenological experience does not typically operate against a backdrop of ineffability, or at least not one of utter or total ineffability. An exploration of the limits of our phenomenological vocabulary – an exploration of the extent to which our phenomenological vocabulary is impoverished, and whether it is in principle possible to overcome this impoverishment – thus seems to be well worth pursuing.

2 Two examples

Our exploration of the impoverishment problem will be helped by having some specific examples of impoverishment before us. For the first example, I’ll focus on one of the philosophical contexts in which philosophers have engaged in efforts at descriptive phenomenology, namely, in philosophical discussion of memory and imagination. This project dates back at least to Hume, who attempted to characterize memory experience in terms of its force and vivacity. In his view, one striking aspect of what it’s like to remember is that the memories present themselves with a certain power, and they are “much more lively and strong” than imaginings (Hume 1739/1985). But while Hume might be right that many memories are more forceful and vivid than many imaginings, this doesn’t seem to be true as a general matter. To give just one example, consider a parent whose son is about to start kindergarten. The parent barely remembers her own first day of kindergarten. The memory that she forms is hazy and undetailed. When she imagines her son’s first day, however, the imagining is clear and fully detailed. Cases such as this show why philosophers have long rejected Hume’s descriptive phenomenology as inaccurate. That said, there’s no consensus about how to do better. Contemporary discussions of the phenomenology of memory point to various feelings – the feeling of not making it up, the feeling of pastness, the feeling of familiarity – that often seem associated with memory. But none of them seems to do justice to the subjective character in question.Footnote 4

Consider the feeling of pastness, for example. Jordi Fernández suggests that the phenomenology of memory seems to involve temporal awareness; when one remembers an event, like entering one’s kindergarten classroom for the first time, the event seems as if it were taking place in the past (Fernandez, 2008, 335–336). Mohan Matthen makes a similar point: “[W]hen I remember eating lunch yesterday, I have an experience that duplicates some of the imagistic features and affective accompaniments of that lunch experience. However, the memory-experience presents itself as about the past – it has, so to say, a ‘feeling of pastness’” (Matthen 2010, 8). But other philosophers have disagreed with these claims and, importantly for our purposes here, their disagreement arises at least in part from the fact that they don’t feel like they have any sort of handle on what the feeling of pastness really is. They don’t know what it’s supposed to pick out and, accordingly, they don’t take the description to be at all evocative of what memory feels like (see, e.g., Byrne, 2010, 23; Debus, 2016, 138).

Granted, some of these philosophical attempts (including Hume’s) are not simply efforts to provide a description of the phenomenology of memory but efforts at providing a uniquely identifying description, i.e., they are attempts to identify aspects of the phenomenology of remembering that differentiates it from the phenomenology of imagining and/or the phenomenology of perceiving. But while providing a unique description of a given phenomenal experience might be a slightly different problem from simply describing the phenomenology of that experience, the two problems go hand in hand. If we could adequately describe what remembering feels like, for example, and we could likewise adequately describe what imagining feels like, then it would presumably be much easier to capture the phenomenological differences between them.

Echoes of the difficulties that have arisen when philosophers have tried to capture the phenomenology of memory are present when we turn to descriptions offered by ordinary folks. Consider one illustrative response that appears in a Quora thread about what remembering feels like:

The best way I can describe it is like plunging into thick water. I don’t know, I always feel my brain diving into its own juices when I summon memories. A part of my brain just pulsates into itself, helping to bring back the menagerie of certain times or images in my life. For me, the physical feeling of remembering isn’t so memorable. I don’t even know how to describe it accurately, just that it vaguely resembles an ocean.Footnote 5

Other posts on the thread reveal a similar struggle to describe the phenomenal character of remembering. To give one more example, consider this passage from user GB who describes the sensation of remembering as “a little like opening the doors within a house where some doors have deteriorated, are ill at ease on their hinges, and suffering from rust.” We also see evidence of the impoverishment problem in a Reddit thread about what remembering feels like:

When I remember stuff, it feels like looking at a picture. This picture is old and has scratches in some places. So much that I wouldn’t be able to tell what even was there before. But some places on the picture are so clear you can see them perfectly. I can kind of tell what’s happening on the picture, but it’s hard to pinpoint any details. And I can’t really imagine what’s happening outside the picture: just what’s in it.Footnote 6

The impoverishment problem is perhaps even more dramatically evidenced when it comes to the case of pain than it is with respect to memory – perhaps because we are more often in situations that require us to describe our pain experiences than we’re in situations that require us to describe our remembering experiences. Empirical studies of qualitative pain assessment frequently mention the difficulty of description. We see this across a many different medical conditions, from rheumatoid arthritis (Bergstrom et al.,2017, 470) to multiple sclerosis (Harrison et al.,2015). In one study about the challenges of communicating the pain involved in endometriosis, 85% of the 131 women claimed to have experienced difficulties describing their pain during discussions with medical personnel prior to their diagnoses (Bullo, 2020). A different qualitative study of how older adults describe chronic pain evidences similar difficulties:

When asked to describe his pain, one man replied, ‘I feel so stupid because I can’t give a proper answer.’ He went on to express his embarrassment about this: ‘I can’t really explain to you how embarrassed I feel because really there isn’t words to explain… I don’t have the jargon to explain what I’m feeling.’ (Clarke et al.,2012, 4).

These remarks by study participants echo the sentiments once expressed by Virginia Woolf in an essay discussing illness:

To hinder the description of illness in literature, there is the poverty of the language. English, which can express the thoughts of Hamlet and the tragedy of Lear, has no words for the shiver and the headache. It has all grown one way. The merest schoolgirl, when she falls in love, has Shakespeare or Keats to speak her mind for her; but let a sufferer try to describe a pain in his head to a doctor and language at once runs dry…” (Woolf, 1993).

In fact, as is perhaps hinted at in this passage from Woolf, our phenomenological vocabulary seems especially impoverished when it comes to pain. Elaine Scarry, for example, suggests that “pain does not simply resist language but actively destroys it” (Scarry, 1985, 4). In her view, pain’s “resistance to language is not simply one of its incidental or accidental attributes but is essential to what it is” (Scarry, 1985, 5). Interestingly, however, she goes on to discuss ways that humans have worked “to invent linguistic structures that will reach and accommodate this area of experience normally so inaccessible to language” (Scarry, 1985, 6). As this suggests, even if the phenomenal feel of pain (or other phenomenal experiences) cannot be fully captured in language, there might nevertheless be some ways that we can enrich our phenomenological vocabulary. I will return to this point in Part IV.

Before moving on, it’s worth noting that there might well be important differences in the kind and degree of impoverishment to which the two examples that we considered in this section give rise, and further differences might well surface were other examples to be considered. We might also think that explanations for the impoverishment will vary depending on the type of experience under consideration. In this way, the impoverishment problem may not be fully uniform across all phenomenological contexts. That said, I think the core phenomenon is likely to be similar enough from context to context that we can treat it as a single phenomenon for the purpose of this discussion.

3 Has the problem been correctly identified?

The discussion of Part II has presented us with two different examples where our phenomenological vocabulary appears inadequate for capturing the phenomenal feel of a given experiences. But even with these examples before us, I suspect that some might be skeptical that there is really as much of a problem here as I’ve suggested. For example, one possible reaction might arise owing to my narrow attention on how things can be described in English. One might worry that the impoverishment problem doesn’t generalize to other natural languages, or even if it does, that the problem may seem especially acute only because I’m focused on English. Other natural languages do better.

As support for this response, one might point to languages that have a finer-grained vocabulary of emotion terms than English does. For example, German has many words for emotions that English lacks such as Weltschmerz, a term that can be loosely translated as “world pain.” Weltschmerz is used to characterize an inexplicable feeling of melancholy that arises even when nothing in particular is wrong.Footnote 7 It’s not just German that has these more refined emotional terms. Consider the Norwegian word forelsket. Used to describe the euphoric feelings when one is first falling in love, it picks out what we might loosely refer to as an electric charge or gut-rush.

Operating within a language where terms like these were readily available would help us to make some progress of the impoverishment problem. But it’s not clear that this progress would take us very far. Suppose that we were to add these German and Norwegian terms to English, just as we’ve done in the past with words like Schadenfreude. While the names allow us to refer to what we’re feeling, and they tell us something about the circumstances in which the feelings arise, the names themselves don’t really communicate what the experience is like – just as with words like “pain.” So enriching English with these terms from other languages doesn’t itself seem to solve the problem. Ultimately, the problem isn’t one about English but one that runs deeper.Footnote 8

But now I envision another possible response to what I’ve said so far. One might worry that the problem isn’t really one of describing what it’s like to have a certain phenomenal experience; the problem is one of introspecting that experience in the first place. Insofar as we’re unable to adequately describe the phenomenal character of our experiences, this inability owes to our inability to adequately introspect the phenomenal character of our experiences. Here one might offer an instructive analogy. Someone driving on the freeway might be unable to describe the car that just passed them. But this inability doesn’t have anything to do with a deficiency in the expressive power of language. It’s not that they don’t have an adequate vocabulary to describe how the car looks; it’s simply that it went by too quickly for them to get a good look at it.

To some extent, it seems right that if we were to get better at introspecting our phenomenal states then that improvement would enable us to do a better job describing them, just like if we were to get a better look at the car passing us on the freeway we would be able to do a better job of describing it. Improving our introspective capacities will undoubtedly play a role in addressing the problem of impoverishment. But it’s hard to see how introspective improvement on its own would solve the problem altogether. First, note that the indescribability of our phenomenal experiences arises even in cases that are very phenomenally impressive (e.g., pain), and even where our introspective efforts are focused and unhurried. The experiences that elude description are not simply those on the periphery or those that are fleeting. Rather, the impoverishment problem arises even in cases of experiences that are central and lasting.

At this point, however, one might point to arguments about the unreliability of introspection. Philosophers like Daniel Dennett (1991) and Eric Schwitzgebel (Schwitzgebel, 2008) have presented a series of examples that lead them to conclude that introspection is considerably more unreliable than we recognize. Though many of their examples involve introspection in what we might think of as fringe cases – introspective efforts at the periphery of our visual field, for example – some of their examples involve more central cases of introspection. If their conclusion is right, then that might seem to suggest that the impoverishment problem is simply a symptom of the problem about introspective unreliability.

A full treatment of this response would probably require us to look more closely at the arguments for the unreliability of introspection and diagnose whether, and to what extent, they connect to the cases where our vocabulary is impoverished. But even without this closer look, I think we can still head off this kind of redescription of the impoverishment problem. Arguments about the unreliability of introspection usually point to cases where introspection leads us astray. We mistakenly think we’re introspecting state S when in fact we’re not in state S but rather S*. But notice that our phenomenological vocabulary can be as impoverished with respect to S as it is with respect to S*. Relatedly, it’s also worth noting that arguments for the unreliability of introspection typically allow that we are able to introspect correctly in at least some kinds of cases. The introspective failure is not across the board. Insofar as the impoverishment problem applies even in those cases, that is, even in cases where it’s granted that introspection has not led us astray, then we can see that the impoverishment problem does not really reduce to a problem about introspection.

As the discussion of this section suggests, the impoverishment problem cannot be easily explained away or reduced to other familiar philosophical problems. Rather, the impoverishment problem is both real and robust. But is there anything we can do to resolve, or at least, diminish the problem? I take up this question in the next section.

4 The way forward

Having recognized the impoverishment of our phenomenal vocabulary, it’s worth noting that it cannot be dismissed as an idle concern. First, it clearly has an impact in a variety of practical contexts. Perhaps the most obvious such context is a medical setting where a physician needs to diagnose and treat pain or other sensations. The inability of a patient to adequately describe and thereby communicate what they are experiencing makes it difficult for a physician to treat them properly. This has proved evident in a number of different medical contexts. For example, in the study of endometriosis discussed above, the inability of patients to communicate their pain seemed to lead to delayed diagnosis of their condition. Or, to take another example, consider the phenomenon of breathlessness. This sensation is often experienced by individuals living with chronic respiratory or heart conditions. One way that this sensation is measured is via a respiratory questionnaire called the Multidimensional Dyspnoea Profile (MDP). In a study aiming to assess the effectiveness of the MDP, a team of researchers discovered that many of the questions posed difficulties for the participants, who were unable to map their lived experience onto the framework used (Malpass et al., 2022). For example, one set of questions on the MDP asked participants to rate how strongly this statement applied when they were doing various activities: “I am not getting enough air, I am smothering, I am hungry for air.” As the research study revealed, this vocabulary was not useful for many participants. One participant summarized the difficulty: “Smothering to me means you’ve got something over your face, over your mouth.… So the question, I don’t really understand the question” (Malpass et al., 2022, 67). As should be obvious, if patients don’t understand the questions on the questionnaire being used to measure their symptoms of breathlessness, devising an adequate treatment plan will prove difficult.

But, over and above the practical import of the impoverishment problem, I contend that it also matters for philosophy. In arguing for the need to develop what he calls analytic phenomenology, Charles Siewert nicely expresses the philosophical import of a problem that seems closely related to the one I’ve been discussing in this paper, namely, that “inadequate or divergent understandings of the terms we use to express questions about the mind hinder our ability to answer them” (2016, 96). As he goes on to describe his project: “the point is to develop concepts that will best serve us in addressing philosophical questions about the mind” (2016, 97). On Siewert’s view, such concepts are needed in order to do philosophical work. For example, consider debates about the existence of cognitive phenomenology, that is, debates about whether there is a distinctive and proprietary phenomenology to thinking a thought, over and above the kind of phenomenology that comes from associated mental imagery. This debate serves as one telling example of a context in which we are hindered by a lack of adequate concepts, for example, an adequate grasp of the concept phenomenology itself. Attempts to refine our phenomenological vocabulary could play a crucial role in helping to develop this concept and many others of the sort that Siewert has in mind, thereby positioning us better to address a number of key debates in philosophy of mind.

So how do we go about achieving the needed refinements? This question is a quite difficult one, but I do think it’s possible to suggest some ways that we can make progress in addressing the impoverishment problem. I’ll here make three inter-related suggestions.

First, we should start by thinking more carefully about the phenomenological vocabulary that we already have and work to develop systematizations or categorizations of it. In doing this, we might take a lesson from some of the practical contexts in which the impoverishment problem arises. For example, some efforts at systematization and categorization have already taken place in the medical context. In one study, subjects were asked to classify 102 descriptors of different aspects of pain (Melzack & Torgenson, 1971).Footnote 9 Efforts such as these led to better questionnaires and diagnostic tools and have thereby improved the ability of doctors to communicate with their patients about pain – though it is clear that considerably more work needs to be done.

Second, we need to recognize that phenomenal experience is multi-layered and multi-dimensional, and this fact needs to be reflected in our phenomenal descriptors. Categorization systems like the one about pain just noted help to tease apart some of these layers and dimensions, and this points to the fact that when it comes to phenomenology we may often be working at too coarse a level of description. In various areas of inquiry, both philosophical and beyond, progress has come when we were finally able to see beyond what had long looked to us as atomic, as indivisible. It is thus not at all surprising that a similar point would apply to descriptive phenomenology as well. The problem is not just that we don’t at present have the right words to describe phenomenological properties but that we haven’t quite homed in on the properties that we should be aiming to describe. (Though I earlier said that the impoverishment problem does not reduce to a problem about introspection, this point shows that progress on the former may indeed require progress on the latter.) In particular, if we were able to identify some phenomenological simples – if we were able to break down complex phenomenological experiences to their atomic parts – we might find ourselves better able to communicate to one another about them.

Here it’s useful to consider the fact that trained musicians can look at an unfamiliar music score and imagine correctly what it would be like to hear the piece being played. The reason that they’re able to do this, as noted by David Lewis in discussing sightreading, is that “new music isn’t altogether new – the big new experience is a rearrangement of lots of little old experiences” (Lewis 1999, 265). Even these “little old experiences” that the sightreader draws on are probably themselves complex. But the point helps us to see how we go about breaking down a complex experience into simpler parts. A similar point arises from a different musical example – namely, the fact that trained musicians can often identify the individual notes of a chord that they’re hearing for the first time. To an untrained musician, the chord presents itself in auditory experience as an undifferentiated whole, but there is indeed a structure there that can be discerned by someone who knows what to listen for. Something similar seems true of some types of taste experience, as when a wine connoisseur can identify a complex flavor profile that is not discernible to the novice wine drinker. Perhaps, then, there is a similar kind of structure in other kinds of phenomenological experiences as well, even ones that do not initially seem to us to have any structure.Footnote 10 Insofar as we are able to discern phenomenological structure, we’d then be able to home in on phenomenological simples – and it seems plausible that it might be easier to describe these simpler phenomenological components than it is to describe the experience as whole. At the very least, we would get a linguistic toehold that could be leveraged in further efforts to refine our phenomenological vocabulary.

But now a worry arises. Acquiring the ability to discern phenomenological structure may change the nature of the phenomenal experience that one is undergoing. When a musician or an oenophile learns to differentiate phenomenological parts in what previously seemed to be an undifferentiated sound or taste experience, they might be better described as imposing structure on their experience rather than identifying a structural aspect that was present all along. More generally, we might worry that any conceptual apparatus we use to help us describe our experience might end up changing it.Footnote 11

On one understanding of this worry, the impoverishment problem becomes in principle unsolvable: Experiences cannot successfully be subjectively described, since the very act of description changes the experience. But I’m inclined to think that we can avoid this alarmist result. Insofar as the descriptions adequately capture the nature of our current experiences, they will provide us with a subjective vocabulary for these (newly described) experiences – and so with respect to these (newly described) experiences, at least, our phenomenological vocabulary will no longer be so impoverished.

A third way that we might make progress in addressing the impoverishment problem comes by looking for other ways of getting linguistic toeholds that could be likewise leveraged. One possibility arises from the use of metaphorical expression.Footnote 12 It’s interesting to note that when people find themselves unable to find words to express their experiences, they often call upon metaphors. Participants in many of the pain studies discussed above often used metaphors when asked to describe what their pain is like. To give just one example, when the participant who was embarrassed about being unable to describe his rheumatic pain was further pressed, he ended up reaching for a metaphor in an attempt to find some way to answer the question:

‘It’s a very difficult thing to describe to you. I used to work in the dairy industry and we used refrigeration quite a lot, and if you listen to refrigeration going through pipes, that’s exactly the same feeling as I have in my feet; it’s a sort of a bubbling.’ (Clarke et al., 2012, 4).

We also see metaphors used in explanations of emotion terms from other languages. Recall, for example, that the Norwegian word forelsket was described as an “electric” feeling.

Metaphors often prove useful in other contexts where description proves difficult, such as the description of aesthetic properties. In a discussion of the elusiveness of music, Nick Zangwill (2011) has gone so far as to claim that metaphor is the only way that the aesthetic properties of music can be captured. In his view, it is impossible to capture such properties in a substantive way by means of literal description. This leads him to what he calls the essential metaphor thesis. It is not just that descriptions of music usually are metaphorical but that they must be.

In advocating for the use of metaphors to help us make progress on the impoverishment problem, I do not mean to endorse the strong claim that metaphors are essential for a solution. As the first two points of this section suggest, there are other possible ways forward. Thus I mean only to be making a weaker claim. Perhaps metaphors will not prove useful in all contexts of impoverishment – and particularly so if, as noted earlier, the impoverishment problem is not uniform. But in at least some phenomenological contexts, I suspect that a reliance on metaphors can help us to make important progress in addressing the impoverishment problem.

What accounts for this progress? One reason is that, as metaphors become more familiar, they gain expressive power.Footnote 13 Presumably this is why, as Virginia Woolf noted, we feel like we are better positioned to describe our feelings of love than we are with respect to describing our experiences of pain. A long literary tradition has provided us with metaphors relating to love that enable us to communicate effectively with one another about how a certain sensation of love feels. Insofar as metaphors help us to capture experiences that we find difficult to capture, it looks like an important step towards reducing the impoverishment of our phenomenal vocabulary would be to develop better metaphors.

Interestingly, some philosophers have rejected this suggestion. In a discussion aiming to describe the phenomenal character of memory, for example, Fabrice Teroni has claimed that “The main challenge posed by memory phenomenology is to explain in non-metaphorical terms how previously acquired information makes itself manifest as such” (Teroni, 2017). But Teroni offers us no reason to think that, in principle, metaphorical language cannot be enough. Absent any such reason, one might think that the proof will be in the pudding. Moreover, the fact that individuals do so often reach for metaphorical language in describing their experiences seems to suggest that such language – if adequately developed in a way that enables interpersonal understanding – might give us greater communicative power. Perhaps one might worry that, in reaching for metaphorical language, the individuals are doing so less as an attempt to offer a description of their phenomenology and more as a theoretical explanation of the phenomenon in question. But this just means that we need to take care when relying on metaphor, and we need to be aware that not all the metaphorical language proposed will be effective in serving our descriptive needs.

The task of refining our phenomenological vocabulary is a difficult one. It will take work. Moreover, this work probably needs to be done in concert with others, as we need to test the vocabulary we come up with to assess its communicative potential. The three suggestions here will undoubtedly need to be supplemented with further suggestions in the course of the refinement project, but it’s my hope that they provide us with a useful starting point and a road map for how to approach the difficult work ahead.

5 Concluding remarks

As we come to the end of our discussion, it will be useful to return to where we started, namely, to Nagel. Recall that his challenge to us was “to form new concepts and devise a new method - an objective phenomenology not dependent on empathy or the imagination.” Of course, Nagel was looking for concepts and methods that would be adequate to communicate about phenomenology across the kinds of divides that separate humans from bats, or blind people from sighted people. But, as I suggested earlier, the problem is broader than that. Thinking about the impoverishment problem as a problem about (or only about) objective phenomenology isn’t quite right. We also need to do a better job with respect to subjective phenomenology. Doing so won’t enable us to know what it’s like to be a bat. But the kind of work needed to enrich our subjective phenomenological vocabulary may well put us in a better position to tackle the problem of developing an objective phenomenological vocabulary.

Doing so also has the potential to enable us to make progress on a wide number of other philosophical problems connected to the mind – from what Mary can know in her black-and-white room, to how we should answer Molyneux’s question about whether someone blind from birth whose sight is restored could differentiate cubes and spheres solely by sight, to whether robots and other machines have qualitative states. But it also has the potential to enable us to make progress on issues in other philosophical subfields. To give one obvious example, it can help to contribute to discussions of empathy.

Given all this potential – the potential to shed light on all of these theoretical matters – I would like to close with one cautionary note. In undertaking the enterprise of developing our phenomenological vocabulary, I’d urge that we take care to be sure that this phenomenological work precedes theory. Unless we do so, there is a danger that our vocabulary becomes more prescriptive than descriptive.

Consider, for example, the fact that discussions of the phenomenology of memory often take place against certain background assumptions about memory – e.g., assumptions about memory’s epistemic role. By imposing these antecedent constraints on our phenomenological investigations, however, we unreasonably tie our hands. After all, we might well be mistaken about memory’s epistemic role, or we might be mistaken that memory’s epistemic role should be able to be grounded in phenomenological facts. If we are to have any hope of achieving an adequate phenomenological vocabulary – and if that vocabulary is to have any hope of contributing to philosophical progress – then it seems important to let our phenomenological efforts guide theory and not the other way around.