Sentientism Still Under Threat

François Kammerer[1]


Reply to Dung

Abstract: In ‘Ethics Without Sentience: Facing Up to the Probable Insignificance of Phenomenal Consciousness’ (Kammerer, 2022), I argued that phenomenal consciousness is probably normatively insignificant, and does not play a significant normative role. In ‘Preserving the Normative Significance of Sentience’ (Dung, 2024), Leonard Dung challenges my reasoning and defends sentientism about value and moral status against my arguments. Here I respond to Dung’s criticism, pointing out three flaws in his reply. My conclusion is that the view that phenomenal consciousness is distinctively significant is still very much under threat.

1. Introduction

In ‘Ethics Without Sentience: Facing Up to the Probable Insignifi­cance of Phenomenal Consciousness’ (Kammerer, 2022), I argued that phenomenal consciousness is probably normatively insignificant: it does not play the distinctive role it is often believed to play when it comes to value and moral status. My argument began with two pre­suppositions. First, our minds are material. Second, our introspection of phenomenal states mischaracterizes its targets. Based on these pre­suppositions, I built an argument against the normative significance of phenomenal consciousness. I notably used two lines of reasoning. First, a rebutting argument against the normative significance of phenomenality (the ‘argument from indeterminacy’). Second, an undercutting argument against this same view (the ‘argument from justification’).

In ‘Preserving the Normative Significance of Sentience’ (Dung, 2024), Leonard Dung responds to my article. He defends what he terms ‘sentientism’ against my criticisms, by challenging key premises in both lines of reasoning.

Dung skilfully reconstructed my arguments and eloquently engaged with them. However, his reply contains three flaws that prevent his objections from being effective. ‘Sentientism’ remains under threat.

I begin by revisiting my arguments (§2). Afterward, each of the next three sections addresses one of the three flaws of Dung’s response. I begin (§3) by delving into Dung’s interpretation of sentientism. I show that the form of sentientism Dung defends is notably weaker than the thesis I discuss in my own article, which renders Dung’s response inappropriate. Next (§4), I assess Dung’s reaction to the argument from indeterminacy. I show that it doesn’t do much to allay the related concerns. Lastly (§5), I evaluate Dung’s response to the argument from justification. I contend that it fails to neutralize the argument.

2. Against the Normative Significance
of Phenomenal Consciousness

In the article criticized by Dung, I contend that phenomenal con­sciousness is probably normatively insignificant, ‘in the sense that it is probably not particularly normatively significant’ (Kammerer, 2022, p. 187). More specifically, I assert that phenomenal states are not likely to play a distinctive role in grounding value — relative to their closest non-phenomenal counterparts. I also assert that the capacity to enter phenomenal states — sentience — is unlikely to play a distinctive role in grounding moral status — relative to the capacity to enter the nearest non-phenomenal states. Both in my article and here, ‘the normative significance of phenomenal consciousness’ is meant to convey the purported notion that phenomenal states and sentience play such a distinctive role in grounding value and moral status. My objective was to show that phenomenal consciousness is not norma­tively significant in this sense, even if it is widely believed to be.

My argument relied on two presuppositions (ibid., pp. 187–8). First, metaphysical materialism (or physicalism): minds are nothing over and above ordinary material processes. Second, introspective mis­characterization: when it comes to phenomenal consciousness, things are not as they introspectively seem. For instance, it introspectively seems that we enter phenomenal states distinct from ordinary material processes (e.g. primitive, immediately known, non-physical states, etc.), but in reality, we do not.

These presuppositions lead us to embrace one of two views. We could endorse strong illusionism: the view that phenomenal con­sciousness does not exist but seems to (Dennett, 2016; Frankish, 2016; Kammerer, 2021). Alternatively, we could endorse reductive physical­ism combined with a dose of weak illusionism: phenomenal con­sciousness exists, but is quite distinct from how it introspectively appears.[2] My stance is that both positions imply that phenomenal con­sciousness is probably normatively insignificant.

Consider first that strong illusionism is true. Phenomenal conscious­ness, being non-existent, cannot ground value and moral status in reality. One might argue that, while phenomenal consciousness does not exist under strong illusionism, quasi-phenomenal consciousness does. Quasi-phenomenal consciousness (Frankish, 2016, pp. 15–16) is the set of real, non-phenomenal mental states, usually mischaracter­ized as phenomenal (e.g. in introspection). At first glance, it might seem that, for a strong illusionist, quasi-phenomenal consciousness could still be distinctively normatively significant — play the role usually ascribed to phenomenal consciousness. This would maintain a viewpoint akin to the one I criticize. I will expand on this shortly.

Second, suppose reductive physicalism combined with weak illusionism is true. In this case, we have two arguments against the normative significance of phenomenality.

Before detailing these arguments, I wish to note that, while I do not elaborate on the reasons here, these arguments could also apply to quasi-phenomenal consciousness if we adopt a strong illusionist stand­point. Hence, they also support the normative insignificance of quasi-phenomenality. More generally, in the upcoming discussion, I present Dung’s objection as well as my response within a weak illusionist and reductive physicalist framework (bearing on phenomenality, not quasi-phenomenality). However, this entire discussion could be recast to bear on quasi-phenomenal consciousness if one leans towards a strong illusionist perspective.[3]

The first of these arguments is a rebutting argument: the argument from indeterminacy.[4] Reductive physicalism combined with weak illusionism suggests that whether or not various beings (notably non-human beings like non-human animals and artificial systems) possess phenomenal consciousness will be indeterminate in many cases. There will be no fact of the matter as to whether or not their mental states are phenomenal (I explain why later, in Section 4). But for phenomenal consciousness to be normatively significant, it must be determinate in most cases. This gives us a reason to deny its normative significance.

The second argument is an undercutting argument: the argument from justification. Reductive physicalism combined with weak illu­sionism implies that phenomenal consciousness is quite different from how it introspectively appears. This divergence undermines our intro­spective justification for the normative significance of phenomenal consciousness. Furthermore, if our introspective representations of phenomenal consciousness mischaracterize their referents, then the phenomenal concepts derived from these introspective representations are also mischaracterizing. This, in turn, undercuts our intuitive a priori justification for the normative significance of phenomenality. All this undermines our justification to believe that phenomenality is normatively significant.[5]

3. Sentientism: Weak, Strong, Distinctive

Dung defends ‘sentientism’ against my objections. More precisely, he defends what he terms ‘weak sentientism’: the view that ‘there are some cases where [phenomenal] consciousness is normatively relevant’. On Dung’s definition, this means the following: there are some instances where phenomenal consciousness ‘intrinsically contri­butes to moral status’ or ‘intrinsically contributes to intrinsic value’ (Dung, 2024, p. 10). ‘Intrinsically contributes’ here means contri­buting non-instrumentally to ground (or constitute, etc.) value and moral status. Dung differentiates this view from ‘strong sentientism’, which combines weak sentientism with the claim that only phenom­enal consciousness is normatively relevant. Dung does not attempt to defend strong sentientism.

This characterization makes Dung’s position problematic. The view I targeted in my own article is stronger than weak sentientism, even if it does not go as far as strong sentientism. Indeed, I specifically addressed the view that phenomenal consciousness is particularly significant when it comes to value and moral status — the view that it plays a distinctive role for value and moral status relative to neigh­bouring non-phenomenal processes. Let’s call this view ‘distinctive sentientism’. One could accept Dung’s weak sentientism while dismissing distinctive sentientism. For instance, this could be done by acknowledging that phenomenal states sometimes play some intrinsic (i.e. non-instrumental) role in grounding intrinsic value, even if this role is not distinctive compared to the role played by neighbouring non-phenomenal states.

While I consider the difference between weak and distinctive sentientism self-evident upon reflection, it might help to flesh it out with an example. Imagine that all phenomenal states are mental states possessing property P: they all intrinsically have P. Moreover, imagine also that some non-phenomenal states, which are otherwise quite similar to phenomenal states, also intrinsically possess property P. (Is there an example of a property that plausibly fits the bill? Maybe the property of being a suitably complex mental state. The specifics are not crucial here.)

Now, let’s add a final supposition. The possession of P is distinctively significant. States with P play a distinctive role for value and moral status — compared to neighbouring states devoid of P.

Given these assumptions, one could comfortably accept Dung’s weak sentientism. Phenomenal states intrinsically contribute to value and moral status — after all, they all intrinsically have P. However, one could simultaneously reject distinctive sentientism, and claim that phenomenal states are not particularly significant, notably because they aren’t more significant than neighbouring non-phenomenal pro­cesses also possessing P.

Compare: sexists believe that human males possess distinctive rights, relative to their closest non-male counterparts — human females. We contend that sexism is flawed and that human males don’t have distinctive rights. Being a human male is not normatively significant, in the sense that it is not particularly so. Yet, this does not imply that human males lack rights — only that these rights are not distinctive compared, for instance, to human females. We believe that, if anything has distinctive rights, it is humans (or more specifically, maybe, entities with human-like mental lives). This also does not suggest that being a human male does not intrinsically contribute to a creature having rights. Arguably, being a human male intrinsically (and not just instrumentally) contributes to an entity possessing rights, but it does not make a distinctive contribution (e.g. contrasted with being human).

So, Dung’s reply is partially misdirected. It defends a view that is weaker than the one I criticized. This is more than mere nitpicking: as I show below (in §5), Dung’s objection to one of my arguments (the argument from justification) becomes notably less compelling once we recognize that the view in question is distinctive sentientism, not weak sentientism.

I also believe we have compelling reasons to focus our discussion on distinctive sentientism (or strong sentientism) rather than weak sentientism. Indeed, underlying these theoretical debates are crucial practical debates about how we should treat various non-human entities. Yet weak sentientism — the view that ‘there are some cases where [phenomenal] consciousness is normatively relevant’ — seems too weak to significantly impact these debates. For instance, weak sentientism in itself has very little to no implication regarding the respective moral statuses of two beings who are nearly identical, save for one being sentient and the other not, or regarding whether a phenomenal pain is worse than a very similar non-phenomenal pain.[6] Distinctive sentientism, on the other hand, firmly suggests that the sentient creature possesses a much higher moral status, and that phenomenal pain is much worse than non-phenomenal pain. I contend that, when people discuss sentientism, they often have in mind views precisely making this sort of suggestion. They contemplate something like distinctive sentientism (or perhaps even strong sentientism) — not weak sentientism.

4. Why the Widespread Indeterminacy
of Phenomenality Remains Probable

Dung’s response suffers from another flaw, which can be examined while setting the previous difficulty aside.

Dung’s first objection addresses the argument from determinacy. Dung concedes that, if the presence of phenomenal consciousness is indeterminate in many cases, phenomenal consciousness cannot be normatively significant.[7] However, he denies that my presuppositions (materialism and introspective mischaracterization) imply that the presence of phenomenal consciousness is likely to be indeterminate in many cases.

4.1. The original argument for indeterminacy

Why did I assert that the widespread indeterminacy of phenomenal consciousness follows from my presuppositions? Here is the rationale.[8] If phenomenal consciousness is real but significantly differs from how it introspectively appears, it cannot be identified with what­ever satisfies the conditions borne by our introspective representations — or our introspection-based phenomenal concepts. If these repre­sentations refer, it is not in virtue of conditions they specify, but in virtue of certain causal-informational factors — e.g. because they are appropriately caused by (or appropriately co-vary with) certain pro­cesses. These processes, which we can term F-processes, are the referent of our representations of phenomenal states.

These F-processes co-instantiate many properties. For illustration, assume that the global neuronal workspace theory of consciousness (Dehaene and Naccache, 2001) is approximately correct. Therefore, F-processes have certain functional properties: e.g. they consist in globally broadcast representations. They also possess neural properties: e.g. they consist in neural activities in the prefrontal cortex. Furthermore, they possess many other properties, which can be seen as determinate of the properties I just highlighted. F-processes are globally broadcast representations, but they are also, more specifica­lly, globally broadcast representations made accessible (among other things) to a decision-making module and a language module, and so on.

So, when we examine other beings whose physical composition and/or cognitive architecture significantly differ from ours (such as non-human animals or artificial minds), we should expect many of them to exhibit processes that resemble F-processes in some ways but not in others. Some might possess globally broadcast representations, but no cortex. Others might have cortex-based globally broadcast representations that are not accessible to a language module, and so on. As a result, in many instances, it is plausible that whether or not the relevant processes really are phenomenally conscious will remain indeterminate. There will not be a fact of the matter as to whether they are sufficiently like F-processes, or like F-processes in the right way. Their categorization as phenomenally conscious will boil down to a semantic decision.

This consequence might be avoided if our introspective representa­tions determinately referred to those F-processes as bearers of some single specific property (which we can label Q). In this scenario, for most (and, ideally, all) creatures, they determinately have (or lack) processes with Q. Therefore, they determinately have (or lack) phenomenal consciousness.

Is this scenario plausible? Something like this would happen if all (or most) F-processes instantiated a single salient specific natural-kind property (supporting a vast range of inductions), which would serve as a reference magnet for our introspective representations, thereby assuming the role of Q. Yet, we have reasons to suspect that F-processes systematically co-instantiate various natural-kind properties (a minima, functional properties and neural properties) and that none is sufficiently uniquely salient to be the single determinate referent of our representations (Birch, 2022). The hope that our introspective representations of phenomenal states latch onto one single salient specific natural-kind property Q appears faint, though it is not entirely inexistent. Hence, the widespread indeterminacy of phenomenal con­sciousness is likely.

4.2. Dung’s objection

Dung challenges this line of thought. He argues that it only holds if one assumes that phenomenal consciousness cannot be equated with whatever satisfies the conditions borne by some of our representations of phenomenal consciousness. This, in turn, for Dung, presupposes that type-A physicalism is false. For Dung, if type-A physicalism is true, our concept of phenomenal consciousness does refer to the property corresponding to the conditions borne by our concept, thereby sidestepping indeterminacy. To be more precise, in the type-A physicalist view Dung considers, our concept of phenomenal con­sciousness comes with an associated conception. This conception con­sists in a set of conditions serving as a reference-fixer. The concept’s referent is the natural-kind property that best fits this conception.

Dung emphasizes that this sort of type-A physicalist view is com­patible with my two presuppositions (materialism and introspective mischaracterization). At least, this is so provided we assume that our concept of phenomenality (and its associated conception) is not derived from our (mischaracterizing) introspection (Dung, 2024, fn. 15).

For Dung, if our concept of phenomenality is such, the widespread indeterminacy of phenomenal consciousness is likely to be avoided. What is this conception associated with our concept of phenomenality supposed to be? First, Dung contends that this conception might encompass a collection of ‘functional truths’ (ibid., p. 21).[9] He pro­poses that these truths will be sufficient to narrow down the reference of our concept of phenomenality to a single specific natural-kind property (ibid., pp. 22–3). Second, Dung suggests that our conception of consciousness also features the very idea that consciousness is not widely indeterminate. Therefore, this conception fixes the reference of the concept in such a way that it refers to a property whose presence is not widely indeterminate.

4.3. Replying to Dung’s objection

Several points need addressing. First, I did not articulate my pre­suppositions through the lens of the distinction between type-A and type-B physicalism (Chalmers, 2002). The difference between the two sets of views is that, for type-A physicalists, there is no epistemic gap between the phenomenal and the physical: there are no (non-trivial) phenomenal truths that cannot be derived from physical truths. Type-B physicalists, on the other hand, admit the existence of an epistemic (but not an ontological) gap: there are phenomenal truths that cannot be derived from physical truths.

I did not dismiss type-A physicalism from the start, since I con­sidered strong illusionism, which is a quintessential type-A physicalist view.[10] What I overlooked are some specific type-A views where our concept of phenomenality is functionally analysable in a manner enabling the deduction of phenomenal truths from physical truths. This is the sort of view Dung has in mind. He claims that these views are compatible with my presupposition of introspective mischaracter­ization, provided our concept of phenomenality is not derived from our introspective representations. However, in the paper, I assumed that ‘our concept of phenomenal consciousness is essentially tied’ to these mischaracterizing ‘introspective representations’ (Kammerer, 2022, p. 199). This is why I did not consider these views. I am willing to acknowledge that this presupposition would have benefited from being made clearer and more forceful — as well as earlier in the article. I did not feel the need to do so because the view that our con­cept of phenomenality does not derive from introspection strikes me as very implausible. The most influential definitions of phenomenal consciousness invariably appeal to introspection and the first-person perspective (Block, 1995, pp. 230–31; Chalmers, 1996, p. 4; Nagel, 1974, p. 436; Searle, 1992, p. 83).

Pushing this primarily dialectical concern aside, let us assume, as Dung proposes, that our concept of phenomenality carries with it an associated conception, separate from the mischaracterization borne by our introspective representations. Let us also assume, in line with Dung, that this conception (a) specifies functional properties of phenomenal consciousness; (b) characterizes phenomenality as always determinately present or absent. Dung’s claim is that, if our concept of phenomenality has these two traits, it refers to a single property whose presence is not widely indeterminate. Is he onto something?

Let us first hone in on (b). Contrary to Dung, I am unconvinced that if our conception of phenomenal consciousness characterizes it as corresponding to a single feature whose presence is determinate, then this makes it so that the concept indeed refers to a single feature whose presence is determinate. Either there really exists one single specific property (whose presence is determinate) which uniquely best corresponds to our conception of consciousness, or there is not — perhaps due to multiple co-instantiated properties (whose presence is determinate) fitting our conception equally well. However, in the latter scenario, our concept of consciousness would not refer deter­minately to one of these properties — even if, for each of these properties, its presence is determinate, which is why it satisfies our conception.

Compare: imagine that I regularly meet with two identical twins, Elias and Ernest, whom I mistakenly believe to be the same person (suppose that they avoid being seen together, and both introduce themselves as ‘Bob’). When I believe I am meeting with Bob, I am with Ernest half the time and with Elias the other half. Undoubtedly, my conception of Bob characterizes him as a single individual whose presence somewhere is determinate. However, this does not mean that my concept of Bob determinately refers to a single specific person whose presence is determinate. Indeed, both Ernest and Elias, two persons whose presence is determinate, are equally good candidate referents for my concept. Arguably, my concept of Bob does not determinately refer only to Ernest, or only to Elias (I set aside the difficult question of knowing the right way of thinking about the reference of my concept).

Now, let us focus on (a). Suppose that we admit, with Dung, that our concept of phenomenality comes with an associated conception detailing various properties of phenomenality. Can this conception narrow down the reference of our concept to some single specific property whose presence is always determinate?

Exploring a hypothetical can provide clarity. Let us assume, again, that the global neuronal workspace theory is roughly correct. Our introspective representations co-vary equally well with at least two similarly salient co-instantiated natural-kind properties — being a globally broadcast representation (G), and being a certain type of neural activation in the frontal cortex (N). Can our conception of phenomenality narrow down the reference of our concept of phenomenality so that it refers to only one of these two?

There is no doubt that some people have beliefs implying that phenomenality is G rather than N, or vice versa. However, the essential question is whether the conception of phenomenality associated with our concept, which should be widely shared and obtainable through something like a priori conceptual analysis (e.g. as intuitive conclusions from thought experiments), genuinely specifies the reference in these types of scenarios.

The problem is that it does not seem to happen. Numerous beliefs about phenomenality which are likely part of our conception of phenomenality (as they are widely shared and highly intuitive) do not help to specify the reference of our concept in this case. For instance, both G and N satisfy the idea that normally functioning human brains are sufficient for phenomenal consciousness, or the idea that a system structured merely as a giant look-up table lacks phenomenal con­sciousness. Some beliefs about phenomenality are highly intuitive and favour N rather than G. For instance, many find it intuitive that Block’s Chinese Nation (Block, 1978) lacks phenomenal conscious­ness, making N (a neural kind) a more suitable match for phenom­enality than G (a functional kind, instantiated by Block’s Chinese Nation). However, other beliefs are also highly intuitive, and favour G rather than N. For instance, many find it intuitive that the gradual replacement of my neurons (and other brain cells) by silicon functional duplicates would not destroy my phenomenal conscious­ness, favouring G over N (Chalmers, 1996, pp. 254–7).

Simply put, allowing our widely-shared, intuitive beliefs about phenomenality to play the role of a reference-fixing conception does not provide much help. Some of these beliefs are consistent, but they do not narrow down the reference to G or N. Those that might do it are inconsistent. We could hypothesize solutions, such as quantifying inconsistent intuitions by degrees of intuitiveness, to ultimately tip the balance for G or N. The problem is that we know (given existing philosophical disagreements) that diverse thinkers evaluate these intuitions differently. Consequently, while some precise patterns of intuition (in which each intuition has a specific force) might decisively appropriately narrow down the reference of our concept of phenom­enality, these precise patterns of intuition are not widely shared. They are idiosyncratic. They are not a plausible part of the conception associated with our concept of phenomenality.

This conclusion should not be surprising. If our conception of phenomenality were truly rich and unequivocal enough to specify the reference of the concept in such cases, there would be considerably less disagreement regarding the nature and distribution of consciousness.[11]

To sum up: if we accept, as Dung himself seems to do, that wide­spread indeterminacy of consciousness is likely if the reference of our concept of phenomenality is fixed merely by causal-informational factors, things do not change much if we allow our widely-shared conception of phenomenality to play a reference-fixing role, as it does in some type-A physicalist views. Dung’s objection to the argument from indeterminacy is unconvincing.

5. Why the Introspection of Value is Undercut

Dung’s second objection tackles my argument from justification. In his interpretation, this argument aims at undercutting the evidential status of introspection when it comes to the normative significance of phenomenal consciousness. As explained above, the argument in fact also challenges the evidential status of certain conceptually-driven intuitions about the normative significance of phenomenal conscious­ness, but we can set this aside for now.

Here is the core of Dung’s objection. We can admit that intro­spective mischaracterization occurs (e.g. phenomenal states intro­spectively seem to be primitive, non-physical, immediately known, even though they are not). However, this does not imply that the evidential status of introspection regarding the value of phenomenal states is undercut.[12]

Why? Dung’s reasoning goes like this: we can recognize that intro­spective mischaracterization takes place, but also believe that phenomenal introspection is reliable in certain respects. For instance, phenomenal introspection appears reliable when determining which phenomenal states we are in. We can introspectively discriminate between experiences of pain and experiences of smelling hot choco­late, and even if we introspectively mischaracterize these states (as non-physical, immediately known, primitive, etc.), the discrimination itself seems reliable.

This suggests that, to assess whether introspective mischaracteriza­tion undercuts our introspective justification regarding value, we need to discern whether value-introspection must rather be ranked with our mischaracterizing forms of introspection or with our reliable forms of introspection. Each perspective yields a different (inductive) conclu­sion concerning the epistemic status of value-introspection.

Dung contends that value-introspection possesses a notable characteristic that sets it apart from mischaracterizing introspection and aligns it more with reliable introspection: it is metaphysically neutral. It does not presuppose any anti-physicalist characterization of consciousness. Its verdicts themselves are not at odds with physical­ism. Therefore, according to Dung, the evidential status of value-introspection remains intact.

I am not convinced that we can conclude that value-introspection escapes undercutting. I believe this for three primary reasons.

First, to understand whether or not undercutting is taking place, we need to be clear about the antecedent epistemic status of value-introspection. As I emphasized in the original paper (and as Dung himself notes), one of the main reasons (if not the main reason) why sentientism seems so compelling is that the value of phenomenal states does not just appear introspectively justified. It seems intro­spectively extremely well-justified — justified in a particularly secure manner. That phenomenal states such as phenomenal pain or pleasure possess striking value, for instance, seems ‘particularly obvious and beyond doubt’ (Kammerer, 2022, p. 183) when compared to the value of non-phenomenal things. The corresponding fact seems to be ‘an obvious datum of introspection’ (Dung, 2024, p. 12). In other words: many people consider value-introspection to hold an exceptionally high epistemic status.

However, this exceptionally high epistemic status is more vulnera­ble than a more modest status — say, an ordinary degree of highly defeasible justification. Recognizing that some systematic intro­spective mischaracterization occurs might already be enough to under­mine it. Even granting that value-introspection is still ‘mostly reliable’ suggests that its epistemic status is inferior to what it might have originally been taken to be. Compare: someone who supposes that the Bible contains grains of historical truth, even if it frequently errs, should not change their belief significantly upon discovering that it contains one specific mistake. However, someone who deemed the Bible the literal and infallible word of God should have their confi­dence shattered under the same situation.

Second, another point deserves mention here, although this was not mentioned in my original article. Suppose we admit that phenomenal introspection mischaracterizes phenomenal states, among other things, as immediately known. If we acknowledge that this introspective mischaracterization occurs, we must admit that phenomenal states are not known in the immediate manner in which they seem to be known. But it is plausible that value-introspection seemed uniquely secure (and its verdicts beyond doubt) because we regarded phenomenal states as immediately known. Accepting that this introspective characterization of phenomenal states as immediately known is mistaken provides a direct reason to suspect that other verdicts of phenomenal introspection — including those of value-introspection — are not exceptionally well-justified. Here, the undermining does not stem from the general fact that introspection is mischaracterizing, but from the more specific fact that it is mischaracterizing in this particu­lar respect. Thus, this undercutting does not depend on value-introspection itself being notably similar to mischaracterizing intro­spection. Compare: if I learn that a message M, once trusted, and which states that person A is uniquely reliable, is in fact mistaken, I should become less trustful of what A tells me — regardless of whether my reasons to doubt M are also directly reasons to distrust A.

Third, Dung’s objection depends on the idea that value-intro­spection is more similar to reliable introspection than mischaracter­izing introspection, because of its metaphysical neutrality. This enables him to conclude that value-introspection itself is reliable. However, there are other significant ways in which value-introspection resembles mischaracterizing introspection more than reliable introspection.

This is clear if we consider that the introspective value-judgments whose epistemic status we examine here are those supporting not merely weak sentientism, but distinctive sentientism — since I argued above that distinctive sentientism is what is really at stake here. Indeed, introspective value-judgments that support distinctive sentientism must assert that certain phenomenal states possess not only just value, but a distinctive sort of value.[13] Consequently, the introspective value-judgments under scrutiny share a salient trait with the judgments produced by our mischaracterizing introspection. Both attribute non-ordinary properties to phenomenal states. By ‘non-ordinary properties’, I mean properties that make them stand out in comparison with non-phenomenal states. In contrast, reliable intro­spection simply differentiates one phenomenal state from another without attributing any non-ordinary properties to any of these states.

While Dung contends that phenomenal introspection has its evidential status undercut only when it is metaphysically loaded (and not when it is metaphysically neutral), one could counter that it is compromised when it attributes distinctive and non-ordinary properties to phenomenal states, and remains unaffected otherwise. Since the relevant value-judgments here — those supporting dis­tinctive sentientism — do attribute non-ordinary, distinctive properties to phenomenal states, this specific type of value-introspection has its evidential status undercut.

This might not solve the issue conclusively. My opponent might engage in some further reference-class tennis. However, combining this with the two previous considerations suggests that our intro­spective justification for believing in the distinctive normative signifi­cance of phenomenal consciousness suffers undercutting to some substantive degree.[14]

6. Conclusion

I have argued against Dung’s defence of the normative significance of phenomenal consciousness, which he developed against the arguments I developed in some previous publication. First, Dung only defends a form of sentientism which is weaker than the one I criticized — and is too weak to deserve being the focus of discussion. Second, Dung’s considerations do not do much to counter the argument from indeter­minacy. Third, his considerations do not defang the undercutting argument from justification, even if the exact degree to which our justification to believe in the (distinctive) value of phenomenality is undercut remains an open question.


Birch, J. (2022) Materialism and the moral status of animals, Philosophical Quarterly, 72 (4), pp. 795–815. doi: 10.1093/pq/pqab072

Block, N. (1978) Troubles with functionalism, Minnesota Studies in the Philoso­phy of Science, 9, pp. 261–325.

Block, N. (1995) On a confusion about a function of consciousness, Brain and Behavioral Sciences, 18, pp. 227–247.

Carruthers, P. (2019) Human and Animal Minds: The Consciousness Questions Laid to Rest, Oxford: Oxford University Press.

Chalmers, D. (1996) The Conscious Mind: In Search of a Fundamental Theory, New York: Oxford University Press.

Chalmers, D. (2002) Consciousness and its place in nature, in Chalmers, D. (ed.) Philosophy of Mind: Classical and Contemporary Readings, New York: Oxford University Press.

Chalmers, D. (2010) The Character of Consciousness, New York: Oxford Uni­versity Press.

Dehaene, S. & Naccache, L. (2001) Towards a cognitive neuroscience of con­sciousness: Basic evidence and a workspace framework, Cognition, 79, pp. 1–37.

Dennett, D. (2016) Illusionism as the obvious default theory of consciousness, Journal of Consciousness Studies, 23 (11–12), pp. 65–72. Reprinted in Frankish, K. (ed.) (2027) Illusionism as a Theory of Consciousness, Exeter: Imprint Academic.

Dung, L. (2024) Preserving the normative significance of sentience, Journal of Consciousness Studies, 31 (1–2), pp. 8–30. doi: 10.53765/20512201.31.1.008

Frankish, K. (2016) Illusionism as a theory of consciousness, Journal of Con­sciousness Studies, 23 (11–12), pp. 11–39. Reprinted in Frankish, K. (ed.) (2017) Illusionism as a Theory of Consciousness, Exeter: Imprint Academic.

Graziano, M. (2019) Rethinking Consciousness: A Scientific Theory of Subjective Experience, New York: W.W. Norton & Company.

Kammerer, F. (2021) The illusion of conscious experience, Synthese, 198, pp. 845–866. doi: 10.1007/s11229-018-02071-y

Kammerer, F. (2022) Ethics without sentience: Facing up to the probable insignifi­cance of phenomenal consciousness, Journal of Consciousness Studies, 29 (3–4), pp. 180–204. doi: 10.53765/20512201.29.3.180

Nagel, T. (1974) What is it like to be a bat?, Philosophical Review, 83 (4), pp. 435–450.

Papineau, D. (2002) Thinking about Consciousness, Oxford: Oxford University Press.

Pereboom, D. (2019) Russellian monism, introspective inaccuracy, and the illusion meta-problem of consciousness, Journal of Consciousness Studies, 26 (9–10), pp. 182–193.

Schwitzgebel, E. (2014) The crazyist metaphysics of mind, Australasian Journal of Philosophy, 92 (4), pp. 665–682. doi: 10.1080/00048402.2014.910675

Searle, J. (1992) The Rediscovery of the Mind, Cambridge, MA: MIT Press.

Paper received October 2023; revised November 2023.

[1]      Archives Henri-Poincaré-PReST (UMR 7117), CNRS, Université de Lorraine, Université de Strasbourg.

[2]      For the definition of weak illusionism, see Frankish (2016, p. 15). I think that all plausi­ble and coherent physicalist theories of consciousness must admit a dose of illusionism — they are weak illusionist views. On some views, this dose of illusionism is explicitly endorsed, and it is sufficiently significant to render the difference between weak and strong illusionism merely verbal (Graziano, 2019, chapter 7; Pereboom, 2019, p. 188).

[3]      I actually prefer the strong illusionist perspective, but I momentarily ignore it to coordinate best with Dung’s response.

[4]      This argument builds on a rich pre-existing literature, most notably on Birch (2022), Carruthers (2019), and Papineau (2002).

[5]      In the original article, I mentioned this distinction between introspective and conceptual justification (Kammerer, 2022, fn. 15 and 16), but I did not stress it as clearly as here.

[6]      In fact, given the way Dung formulates weak sentientism (‘there are some cases where phenomenal consciousness is significant’), the view does not even entail that a sentient creature has some moral status — but I leave this aside.

[7]      Dung admits that technically some form of sentientism could still be saved. It is clear that if phenomenal consciousness is determinately present in at least one case and makes an intrinsic contribution to value or moral status in this one case, this is enough to save weak sentientism. However, Dung calls this view ‘something that proponents of sentientism might want to avoid’, since it would make phenomenal consciousness normatively relevant only in a very small number of cases (Dung, 2024, p. 16).

[8]      Again, this builds notably on Birch (2022), Carruthers (2019), and Papineau (2002).

[9]      Here are the examples that Dung gives of such functional truths: ‘When a being reports the presence of an object, it is usually conscious of the object’; ‘When a being’s cog­nitive architecture is organized as a mere look-up table, it is not conscious’; ‘Having a brain as complex and interconnected as the human brain is normally sufficient for con­sciousness’ (Dung, 2024, p. 21).

[10]    ‘Type-A materialism sometimes takes the form of eliminativism, holding that con­sciousness does not exist and that there are no phenomenal truths’ (Chalmers, 2010, p. 111).

[11]    A related idea is that every coherent and plausible view of consciousness has at least one radically counter-intuitive consequence — see Schwitzgebel (2014) for a con­vincing case.

[12]    As previously, I formulate the theses here in a weak illusionist framework, but all this reasoning could be reformulated as bearing on quasi-phenomenal consciousness if one assumes strong illusionism.

[13]    Whether this value is distinctive for quantitative or qualitative reasons does not matter much here.

[14]    The exact degree is difficult to assess, as I acknowledged in my previous article (Kammerer, 2022, p. 195).

Stay Informed, Stay Inspired

Discover the frontier of academic insights with our newsletter. Imprint Academic brings a world of scholarly discourse right to your inbox. Stay updated on the latest publications, special offers, and the vibrant conversations shaping today’s academic landscape. Your quest for knowledge deserves a companion as dedicated as our newsletter. Sign up now and become a part of a community driven by curiosity and intellectual exploration.
Something went wrong. Please check your entries and try again.