Abstract
It is widely assumed in the bioethical literature that the existence of any absolute moral principle depends on the ability of moral intuition to be in line with metaphysical reality. It is further assumed by many that we have shared access to (at least some) core moral intuitions and shared knowledge by which we can easily identify most (or even all) humans as possessing incontestable moral status. According to these assumptions, debate over moral status only needs to address the status of entities such as non-human animals and human embryos. We have previously argued that Gewirth's Principle of Generic Consistency (PGC) is the supreme principle of morality (indeed, of all practical reasoning) and requires no such assumptions, but its application requires a particular type of moral precaution when identifying whom/what beyond oneself is to be granted moral status. John Coggon argues that our ‘moral precautionary thesis’ is incapable of dealing with the ‘metaphysical and epistemological’ challenges that underpin the problem of other minds, which he therefore dismisses as a ‘mistaken starting point’. In this article, we argue that Coggon has not fully appreciated his own assumptions and what is implied by recognition of the PGC as the categorical imperative.
Introduction
In The Hitchhiker's Guide to the Galaxy, the supercomputer Deep Thought takes 7.5 million years to provide the ‘Answer to the Ultimate Question of Life, the Universe, and Everything’. 1 It turns out that the answer is 42. Deep Thought chides the Magratheans for their bewilderment and dismay by pointing out that they did not understand the question to which 42 is the answer. Our ‘moral argument for the recognition of other minds’, first published in a joint paper, 2 (p42) is similarly bewildering and dismaying to those who have not clearly identified the question to which it is the answer. In this article, we seek to elaborate our focal question and the answer provided by our moral precautionary thesis (hereafter MPT) in reply to John Coggon's thought-provoking critique. 3
We have defended and developed the MPT in several works, both jointly and individually, including in response to criticisms advanced by John Coggon and Søren Holm.4,5 Coggon's latest critique addresses the articulation, defence and application of this thesis in Shaun Pattinson, Law at the Frontiers of Biomedicine: Creating, Enhancing and Extending Human Life (hereafter LFB). 6 Since Coggon's target is the MPT, rather than the other features of that book's thesis or its detailed case studies, we consider it appropriate to provide a joint reply.
We present our response in three parts.
Part One summarises the MPT and situates it in the application of a categorical imperative in light of the limits of what we can know. The specific moral principle we apply is Alan Gewirth's ‘Principle of Generic Consistency’ (PGC). 7 This principle holds that an agent (i.e. a vulnerable being able to choose purposes to pursue) a must recognise that all agents have rights to the generic conditions of agency (which are things that are necessary for any action or successful action in the sense that their absence prevents or lessens the possibility of any action or successful action, regardless of the purposes involved). Furthermore, these ‘generic rights’ are rights under the will conception (will-rights), according to which the agent possessing the right may release others from the correlative duty to act out of respect for the generic interest protected by that right. This has the consequence that only those with such a releasing ability can have these generic rights.
Since we use the above abbreviations throughout, we will repeat them here:
MPT
LFB
PGC
Part Two elaborates on our response, outlined in Part One, to Coggon's central claim that the ‘unverifiable metaphysical assumptions’ that we say are needed to know that another is an agent are also needed to know whether ‘other beings/things’ we observe are more or less probably agents. 3 We argue that Coggon errs in thinking that the MPT must (and purports to) secure degrees of ontological probability of agency by failing to fully appreciate the nature and implications of a categorical imperative given the limits of possible knowledge.
Part Three addresses Coggon's particular concern with the claims that we make for potentiality. We argue that this concern stems from the above error and, we further contend, Coggon has misunderstood the purpose and nature of the analogy made in LFB.
The principle of generic consistency and the limits of knowledge
In Reason and Morality, Alan Gewirth presents an argument for the PGC as the supreme principle of morality. 7 This argument is summarised very briefly in Coggon's article, and we anticipate that his article will be published in the same issue as this reply. It seeks to show that the PGC is ‘dialectically necessary’. It is conducted ‘dialectically’ from the internal perspective of an agent and draws out the implications of premises that cannot be coherently denied within that perspective (‘necessary’ premises and implications).
However, to understand our position, it is necessary for us to make some general remarks about the nature of a categorical imperative and the method that Gewirth uses to argue for the PGC as such.
To begin with, the idea of a categorical imperative is the idea of an imperative that governs what an agent may and ought to do regardless of what the agent wants to do, who the agent is, or the circumstances in which the agent is placed. As such, it is analytic that all agents who understand the concept of a categorical imperative ought to think that they unconditionally ought to comply with such an imperative. In effect, the categorical imperative is ‘Act only in accord with a categorical imperative!’ This has the further consequence that a categorical imperative can only exist (be a genuine imperative for action) if:
agents unconditionally ought to have the concept of a categorical imperative in order to fully understand what it is to be an agent (which is to say that understanding its concept is a strict requirement of agential self-understanding, i.e. dialectically necessary); understanding its concept as dialectically necessary reveals that it has a substantive content; and it is coherent to think that an agent can be motivated to act for the sake of it without any other incentive, i.e. it is not incoherent to think that an agent can care sufficiently about dialectical necessary requirements to treat them as ends in themselves (which will be the case if it cannot be proved that this is impossible).
8
What Gewirth's argument seeks to show is that the PGC is analytically contained within agential self-understanding. The justification for using dialectical necessity as the criterion for a categorical imperative is that it is analytic that agents ought to be able to understand the concept of a hypothetical imperative (such imperatives being created by understanding the necessary means to purposes they choose to pursue as a condition of being able to act), but cannot fully understand this concept without also understanding the concept of a categorical imperative. It is also on this basis that a categorical imperative is a strict requirement of agential self-understanding.
For it to be coherent for an agent to think that the agent can be motivated to act solely out of respect for a categorical imperative, it must be the case that (i) a categorical imperative can coherently be thought to exist and (ii) the powers that are presupposed by thinking that agential self-understanding is possible cannot be explained by anything other than that they are presuppositions of any thought at all. The first of these conditions is satisfied by the fact that the concept of a hypothetical imperative cannot be fully understood without understanding the concept of a categorical imperative, given that it is necessary to understand the concept of a hypothetical imperative to achieve agential self-understanding. The second is satisfied by what we consider to be an absolutely basic fact: that all thought (as against mere feeling) presupposes an ‘I’ and that the idea of an ‘I’ requires use of the powers of thought necessary to have agential self-understanding (because all reasoning has a purpose) in consequence of which nothing that exists outside agential self-understanding can be used to explain them. In this sense, the existence of a categorical imperative presupposes that agents can have free will. Gewirth makes some comments relevant to this effect in Reason and Morality, 7 (p1−21,36,37) but we consider that the fullest account along these lines is given by Kant as we understand him, which is rather different from most interpretations. 9
From this it should be clear, contrary to dominant approaches within bioethics, that Gewirth's justificatory method does not in any way rely on moral intuition. Nor does it rest on any metaphysical assumptions. In fact, it rests on the idea that it cannot be known whether agents have free will in a sense (transcendent free will, a self-causing power) that is incompatible with determinism (according to which nothing can be self-caused) or not, whether God (defined as the Omnipotent Perfectly Good Prime Mover) exists or not, or whether experience can discover the workings of transcendent reality (reality outside of the strict requirements of agential self-understanding as it is in itself). Contrary to what many believe, the existence of a categorical imperative does not depend on the existence of transcendent free will or of God. It rests on the intelligible possibility (as against the ontological possibility) of transcendent free will and (for reasons we cannot explain here) on that of God (which is to say on the fact that neither can be proved not to exist). Correlatively, it does not depend on the falsity of determinism but only on the unknowability (i.e. the unprovability) of determinism. This is simply because a categorical imperative, in being wholly unconditional, is a fact of agential self-understanding by itself, what Kant calls a fact of pure reason, not an empirical fact or of any a priori intuition (moral or metaphysical). 10 (p28,29) To be sure, if transcendent (ontological) free will (albeit unknowably) does exist, then it is a property of the powers of thought that makes agential self-understanding possible whereas, if God exists (again unknowably), God exists whether any agential understanding is possible or not, which is why Anselm's ontological argument for God is invalid. b This is important because if transcendent free will exists it is a property of these powers, and these powers are ones that agents necessarily presuppose they have but cannot explain. So even though they cannot know that other beings have these powers, the fact that they know that they themselves have them means that the fact that they can know empirically that other beings behave as though they have them, gives them a handle on which to interpret the agent-like behaviour of others as signifying the existence of transcendent free will for moral purposes without this securing knowledge of its transcendent existence. The key to this is that because even though they cannot know how it is possible that they themselves have the powers of thought, it remains intelligibly open that others who behave as though they have them actually do and, if so, are actually agents to whom they can attribute transcendent free will for moral purposes (but only for these purposes). This holds despite there being no empirical experience that they can similarly regard as intelligibly signifying the existence of God.
In LFB, the name Orion is given to an individual representing any particular agent. (For the same expository purpose, the other author of this article has used the name Agnes, which is conveniently one letter short of being an anagram of agents.) 8
Now, the particular problem that prompts the MPT is that Gewirth's argument for the PGC operates on the assumption that there are other agents. But agency requires the powers of agential self-understanding (the powers of thought) and, as we have already said, the only being that Orion can know can think is Orion. Another being, for example, Dawn, might behave as though she is an agent, but for Orion to conclude from this that she actually is an agent requires unverifiable metaphysical assumptions about the relation between mind and body. This might seem to leave it open for a sceptic to hold that the PGC has no categorically binding application in practice because Orion cannot know that Dawn is an agent and so is not categorically bound to treat Dawn as an agent.
For this reason, application of the PGC as the categorical imperative requires a solution to the problem of other minds. LFB addresses this using two thought experiments. In the first, Orion interacts with Dawn, who behaves as if she is an agent and is therefore described as an agent-behaver. In the second, Orion interacts with Sirius, who is a newborn baby who does not fully behave as if he is an agent, so is described as a partial agent-behaver.
While none of the inferences Orion draws on the basis of empirical knowledge can establish that either Dawn or Sirius has a mind, adherence to the PGC (as dialectically necessary) is categorical. As such, Orion must avoid acting contrary to the PGC if it is at all possible for him to do so, provided only that he does not thereby disproportionately violate his own generic rights. Since Orion's inferences from what he observes have implications for other agents (if there are any), he must act (i.e. morally ought to act) to ensure that he treats all his intelligibly other-affecting actions as if he knows them to be other-affecting. The only qualifications are those implied by the preconditions of practical precepting (such as the principle that ‘ought’ and ‘may’ implies ‘can’) 11 (p48) and the PGC itself.
Orion can intelligibly treat Dawn as an agent, because he observes that she behaves like one. He must therefore ignore the speculative (metaphysical/ontological) possibility that she is not in fact an agent. Coggon mischaracterises this inference: This is because, the argument goes, of the probability of its being an agent given what it seems to Orion to be, and the great moral offence of not respecting its rights should it be an agent.3
Having drawn this further inference, it is possible to consider the PGC's implications for Orion's actions with regard to Sirius. Sirius does not behave like an agent but displays some of the characteristics and behaviours that enable Orion to treat Dawn as though she is an agent. While Orion cannot intelligibly treat Sirius as though he is an agent (or grant him the generic rights or any other will-rights), it is both possible and intelligible for Orion to guard against mistakenly treating Sirius as a non-agent by acting towards him in ways that would respect his generic rights should he (unknowably) happen to be an agent. Moral precaution here therefore requires Orion to grant Sirius interest-rights tracking his presumed generic interests.
8
(p36)
Let us pause to briefly consider what is involved in Orion identifying an agent-behaver and a partial agent-behaver. Orion identifies Dawn as an agent-behaver by mapping the empirical evidence he has on to the attributes conceptually required to be an agent. Relevant empirical evidence (i.e. observed behaviour and characteristics) can be observed in other beings to lesser degrees. It is this process that enables Orion to distinguish:
those who act like agents (agent-behavers), those he suspects behave like agents but about whom he has genuine and legitimate doubt due to significant physiological differences to his own (disputable agent-behavers), those who only partly act like agent-behavers (partial agent-behavers), and those who do not display any characteristics and behaviour of the type expected to be displayed by an agent (non-agent-behavers).
After applying the MPT to Dawn and Sirius, it is stated in LFB (as is quoted by Coggon) that: A single-variable conflict between respecting a will-right of an agent-behaver (Dawn) and respecting the equivalent interest-right of a partial agent-behaver (such as Sirius) is to be dealt with in the way most likely to comply with the PGC, given the epistemic gap. This is an instance of the ‘criterion of avoidance of more probable harm’.
8
(p36) partial agent-behavers are owed duties in proportion to the degree to which they approach being agent-behavers. Thus, all other variables being equal and standard species expectations applying: the life of an embryo counts for less than the life of a fetus, the life of a fetus counts for less than that of a newborn at full-term and a newborn's life counts for less than the life of a child who has reached the stage of being capable of being treated like an agent.
6
(p36) Insofar as another entity seems as if (behaves as if) it is an agent, we infer a high probability of its being that and act towards it as if we know with certainty that an agent is indeed what it is. Insofar as it is not obviously a moral agent, but might be one, we should not risk infringing our moral obligations, and thus ought to treat it as if it is an agent. To do this, Pattinson stipulates markers of probability; including things such as being a member of the human species, albeit without otherwise exhibiting agent-like behaviour.3
Coggon's central claim and its fallacy
In our reply to objections advanced by Holm and Coggon, we state that Gewirth's argument renders the PGC categorically binding on vulnerable agents in relation to those of their actions that affect other vulnerable agents. It is possible, however, to imagine a sceptical agent ‘A’ accepting this yet denying that the PGC has any practical application on the grounds that A is the only agent in the world.
4
(p259) I feel bound to say that I did not suggest that the problem of other minds was a ‘fictional construct’. The point that I sought to make, and with luck present more clearly here, is that Pattinson's response to the problem does not work given the logic of embracing the ‘epistemic gap’ entailed in the precepts of the problem of other minds.3 The same ‘epistemic gap’ between observing an ‘agent-behaver’ and knowing that it is an ‘agent’ also sits between observing and knowing what, if any, behaviours, characteristics or other observable phenomena give rise to more or less probable agency at all where something does not exhibit itself as an agent. The unknowability that we take of the true existence of other agents is bound up in the same epistemological tangle that gives rise to the unknowability of indicators of possible or more or less probable agency in other beings/things.3
The observable behaviours and characteristics of two partial agent-behavers can differ. In some cases, the observable phenomena associated with one will be closer to those associated with being an agent-behaver than for the other, i.e. there are degrees of such observable phenomena. As stated in Part One, our claim is that, when applying the MPT in a single-variable conflict, priority is to be given to the partial agent-behaver displaying the higher degree of these phenomena/appearing to be closer to being an agent-behaver. That is both because the being closer to being an agent-behaver has more interests that might be generic rights and because fewer or more plausible metaphysical assumptions need to be made to explain why that being is not behaving like an agent than need to be made to explain why the other is not behaving like an agent.
Imagine, for example, a single-variable conflict between a tadpole and a healthy adult frog. Neither display sufficient behaviour or characteristics to be intelligibly treated as an agent; neither are agent-behavers. Both display some of the observable phenomena associated with being an agent-behaver. But there are more such phenomena in relation to the frog than the tadpole. The frog, for example, can be observed to perform a greater range of challenging activities and to possess a more developed brain. No claim has to be made about the relationship between the observable phenomena and metaphysical reality. These are to be treated (under moral precaution) as degrees of approach to being an agent-behaver. But this does not make it any more or less probable that the other is or is not actually an agent.
In our opinion, scepticism about the possibility of a categorical imperative as well as about our moral solution to the problem of other minds is based on the belief that any proposition taken to be self-justifying (which a categorical imperative must be) is arbitrary, with the consequence that there is no such thing as objective knowledge or absolutely rationally necessary beliefs at all, unless there is direct access to transcendent reality, either through sense-experience or a priori intuition of some kind. 8 Despite the orthodoxy of this position, we consider it to be wrong because holding it, like holding any position, requires the ability to think, which requires the ability to doubt, affirm and deny statements, to judge what is, to imagine what might be, etc. But, as we have repeatedly reiterated, having this ability requires the powers of agential self-understanding, which cannot be explained by anything outside the remit of agential self-understanding but are presupposed by the possibility of any thinking. As such, Gewirth's argument is circular because it justifies the PGC as the categorical imperative by requiring agential self-understanding to justify itself, inviting the question ‘Why ought I to care about the strict requirements of agential self-understanding?’, as though it matters that I might not care about violating it enough to deter me from disobeying the PGC. But to think that the fact that Orion might not care whether he acts contrary to the requirements of agential self-understanding is irrelevant to the existence of a categorical imperative. This is simply because any imperative requires it to be possible to disobey it as well as to obey it. Most importantly, this particular circularity is not vicious because the idea that dialectically necessary entailments must be accepted subject to their internal coherence and intelligible practical application cannot be coherently questioned. Ultimately, the mistake made by questioning this is fallacious in presupposing that one can know something without being able to think and to think that one can think without using the powers necessary for any thought to be possible.
It is this property of dialectical necessity that also means that the mere intelligible possibility of a categorical imperative makes belief in a categorical imperative not merely rationally permissible but rationally necessary. It is by the same token that the mere intelligibility of agent-like behaviour evidencing being an agent makes it rationally (i.e. morally) necessary to treat it as doing so without this evidencing it as a fact of what it is in itself.
Coggon's critique of potentiality in the moral precautionary thesis
Coggon quotes a lengthy paragraph from LFB before devoting over a page to analysing and responding to it. We reproduce that paragraph here, with the emphasis added by Coggon but without the mistranscription of its second word: A healthy human embryo or fetus is a partial agent-behaver with a feasible developmental pathway or trajectory for becoming an agent-behaver. It is, in this sense, a potential agent-behaver. But its increased status (over a partial agent-behaver that is equivalent save for the absence of such a developmental trajectory) is not due to treating potentiality as actuality. It is due to its comparatively greater closeness to being an agent-behaver. For this reason, Holm and Coggon display misunderstanding when they criticise this precautionary thesis on the basis that ‘it is not clear that general potential to become an agent has any bearing on something's being an agent’. It is true that potentially being X is not equivalent to being X, whether X is ‘being an agent’ or ‘being an agent-behaver’. But the claim being made is very different: it is that potential to become an agent-behaver – in the sense of a feasible developmental pathway to becoming an agent-behaver in the future – has a bearing on closeness to being an agent-behaver. A germinating acorn is not an oak tree; but it is closer to being an oak tree than a tomato seed is to being an oak tree.
6
(p104) What is in issue is that the question of ‘what has a bearing on closeness to being an agent-behaver’ is not, other than as an ‘unverifiable metaphysical assumption’, something that can be evidenced. Against his own epistemic framing, and his concept of agents being defined by rational capacities rather than physical (less still biological) essence, Pattinson is, in the above passage, just making assertions.3
Secondly, Coggon objects to the reliance on the notion of a developmental trajectory and the analogy italicised in the above passage, which contrasts the relationship between a gestating acorn and an oak tree and a tomato seed and an oak tree. He expresses his objections to (what we will call) the oak analogy from two angles. He starts by arguing that a germinating acorn is no closer to being an oak tree than a tomato seed is to being an oak tree: as a matter of analysis, Pattinson holds that the acorn and the oak tree are distinct things, practically and conceptually (it is not an oak tree we are told; not it does not seem to be an oak tree). What ‘closeness’ should mean at all once we have and know we have distinct things is unclear. We do not need to reason further because we already know they are distinct. But the whole point beneath all this is not about conceptual unclarity within the theory; it is about epistemic uncertainty (‘unverifiable metaphysical assumptions’) in the theory's application. Following the analogy, we would be acting, in our applied ethics, whilst knowing that the ‘partial agent behaver’ is not an agent. And it would not be an exercise in addressing an epistemic gap at all.3
As Coggon recognises, the claim that potentiality/developmental trajectory is a relevant factor when considering partial agent-behavers under the MPT has its origins in our first co-authored publication on this thesis. 2 Before we can respond to its presentation in LFB, we must again begin by addressing some misrepresentations of the MPT and the question to which it is a response.
The purpose of the quoted analogy was to explain the notion of a developmental trajectory that (a) applies to some but not all partial agent-behavers and (b) is to be used when applying the MPT in a conflict between two partial agent-behavers where this is the only relevant distinguishing feature. The analogy's purpose was not justificatory, but explanatory. It should also be borne in mind that the MPT is triggered by Orion's attempt to apply the requirements of the PGC to his genuine and competent attempt to make sense of his sensory experiences. He perceives objects to have the attributes of an acorn/oak tree, partial agent-behaver/agent-behaver, etc., but cannot know whether those attributes signify their real/ontological properties. The epistemic gap to which Coggon avers is simply irrelevant to Orion's identification of an oak tree, agent-behaver or any other entity when making sense of his sensory experiences. His ability to so conceptualise those features of his experience makes no knowledge claims as to the agent-independent existence of either those entities or their classification.
The oak analogy does not (nor was it meant to) directly answer Coggon's request for an explanation as to how Orion is to identify a relevant developmental trajectory or provide the justification as to why such a relevant developmental trajectory can be an applicable factor in the application of the PGC under precaution.
A developmental trajectory is an inductive inference from the observation that the observed entity displays the same attributes as other entities that have later matured into entities with characteristics that the observed entity currently does not display. Orion can inductively predict that a germinating acorn could become an oak tree by a developmental process that he does not inductively predict could lead to a germinating tomato seed becoming an oak tree. This reasoning has a place in the MPT only insofar as sound inductive reasoning reduces the number or implausibility (strictly, the complexity) of the metaphysical assumptions that Orion needs to invoke to make sense of the morally required presumption that the entity in question is an agent by identifying more generic interests that might speculatively be generic rights, which is necessary to think of the being as not inconceivably an agent and make it intelligible for agents to have any duties directly to it.
Now, imagine again, that Orion has a single-variable conflict between two entities, Entity A and Entity B, that differ only in that Entity A is (inductively) predicted to develop into an agent-behaver, whereas Entity B is not so predicted to have this developmental trajectory. Our claim is that it requires fewer or more plausible metaphysical assumptions for Orion to hypothesise (as required by the moral precautionary presumption of agency) that Entity A (when compared to Entity B) is a locked-in agent that is unable to display its agency in a way observable to Orion.
Coggon asks the reader to consider which is closer between a germinating acorn and an oak tree, or a full-fledged ash tree and an oak tree? It almost sounds like a facetious question, but the point is that the arguments simply can go either way, with it anyway remaining the case, per how the example is presented, that neither actually is an oak tree.3
In any case, to directly address Coggon's question, from the perspective of the MPT, an oak tree and an ash tree are equivalent in moral status as partial agent-behavers, as are an acorn and a tomato seed as partial agent-behavers. Contrary to what he suggests, it is not an open question whether an acorn is closer to an oak tree than an ash tree is to an oak tree. An ash tree is clearly closer to an oak tree in moral status than an acorn is to an oak tree.
Conclusion
The MPT is our answer to a particular question. Given that the PGC is postulated by agential self-understanding, and not only can I (any agent) not have knowledge that does not require use of the powers of agential self-understanding but only I can know that I have these powers, this question is as follows: How, for moral purposes, am I to make sense of empirical knowledge about how beings behave?
Much of what we have said is controversial as it is grounded in our understanding of Kant's transcendental philosophy and our interpretation of its implications. Both of these make claims and moves that run counter to commonly accepted views of what Kant holds and its validity. This, however, is not to say that we agree with Kant as we interpret him on everything. However, it has not been possible to spell out our views on this fully here, let alone to justify them. Part of this has been done elsewhere,8,9,13 but it is still work in progress. Our aim has merely been to try to show just what we are claiming in our use of the MPT and to clarify what we mean by the idea of precautionary probability with which it operates.
Footnotes
Declaration of conflicting interests
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The authors received no financial support for the research, authorship, and/or publication of this article.
Data availability statement
Data sharing not applicable to this article as no datasets were generated or analysed during the current study.