Is it rational for me to want to have a child if I cannot know what it will be like to have a child? Laurie Paul, in her book Transformative Experience, argues that this question poses a significant problem for traditional theories of decision-making.
Paul holds that many major life decisions are ones in which we cannot grasp the nature of the outcomes of our choice. One cannot grasp what it is like to have a child or commit themselves to a particular career path. A deaf person cannot know what it is like to receive a cochlear implant and hear sound for the first time. If we cannot grasp what these outcomes are like, we cannot rationally assign a value to them. Paul also believes that such major life decisions can change us in ways that we cannot foresee. The very things one cares about alter as a result of the experience. If this is true, how is one to choose rationally in such circumstances? I do not know how much I value the potential outcomes of my decision, and I do not know whether I should be choosing on the basis of my current preferences or my future (unknown) preferences.
In order to understand the particular problem this poses for decision theory, we must first say a little about Paul’s conception of decision theory. Paul believes rational preferences are those that maximise our expected subjective value, given the probabilities we attach to each outcome obtaining. Probabilities are generally thought to be subjective degrees of belief. That is, they reflect the agent’s own assessment of the likelihood of some event occurring. It is not unnatural for a decision theorist to hold that the appropriateness of subjective degrees of belief are also constrained by objective features of the world. We believe that an outcome has a certain probability because, for example, we have seen this probability obtaining in the long run (in a million flips of a coin, the probability that it leads on Heads starts to tend to 0.5). How do we determine appropriate subjective values? Certainly, some people would hold that any value is acceptable. There ought not to be any rational constraint on desires (though there may be moral or other such constraints). For example, it seems perfectly rationally permissible that I attach greater value to eating an apple than a tomato without any further justificatory reason. However, Paul believes subjective values do need to be constrained. In particular, she thinks that our subjective values must be informed by way of imaginative acquaintance in order to be rationally justified.
Why would being able to imagine an outcome constitute a good rationality constraint on how much we should value it? Importantly, outcomes in Paul’s conception in terms of ‘what it is like’ – they necessary involve, but are not limited to, experiential value. How do we come to know what something is like? We imagine it. How do we successfully imagine it? We use previous experience to guide our imaginative projection. If I have had previous experience of eating apples, I can imaginatively assess what it would be like to eat another fruit that is fairly similar to an apple – say, a pear – and as a result of this imaginative acquaintance I can assign a rationally justified subjective value to it. An issue arises when the fruit in consideration is wildly different to an apple. What if I am to decide whether or not to eat a durian fruit? Durian fruits are notoriously foul-smelling and people’s reactions to their taste differ vastly. Now, my previous experience of eating an apple does not seem sufficient for me to assign a rationally justified value to the experience of eating a durian fruit. If this is case, in the terminology of Paul, we describe the experience of eating a durian fruit as epistemically transformative.
In the case of having a child, not only might I be unable to know what the experience is like, but undergoing the experience might change what I am like. Perhaps prior to having a child, Ann cares about her career and being able to sleep in on Sunday mornings. After having a child, it is conceivable that her preferences drastically change. She would rather give up her job and stay home with the baby and she does not mind waking up early on Sundays to feed her; in fact, she comes to value these aspects of her life. Not only is a scenario such as Ann’s conceivable, but it is often observed in the world. Another problem for Ann in deciding whether or not to have a child, then, is that she does not know whether her preferences will change after undergoing the experience; she does not know which preferences will change; and she does not know by how much. If this is the case, we describe the experience as both epistemically and personally transformative.
The challenge is not as straightforward as: should Ann prioritise the preferences of her current self, who prefers to further her career and remain childless, or her future self, who prefers to be a devoted mother? Rather, the challenge is that the preferences of her current self are simply not comparable to the preferences of her future self. The relevant comparison must be between the agent who results from having or not having the child, but yet who has the preferences of the agent’s current self before making the decision. This is a person who is psychologically accessible to the agent, but does not exist in any physically accessible world as the transformative experience, by definition, changes her preferences.
If we cannot assign subjective values to outcomes and we do not know which values (current or future) are relevant to our decision-making, it seems like we cannot make the decision to have a child rationally. In fact, Paul thinks we can, but the way in which decision theory demands we make the choice appears to be in tension with how we would intuitively like to make the choice. Paul argues that standard decision-theoretic rationality would have us gather statistical data and testimony from friends and family about their well-being after they have had children, and replace our subjective preference with whatever rational preference results from such third-personal data. We might ask, what is wrong with this? Imagine the following scenario: Sally has always wanted a child. She treasures the moments she is able to spend with others’ children. She believes that having a child is fundamentally important to her sense of self-fulfilment and purpose in life.
However, upon conducting research, she comes to learn that an overwhelming proportion of people experience lower subjective well-being after having children. Sally does not know what the relevant psychological similarities are between her and the average respondent in the surveys she reads. This means she cannot adjust the data to discover her own value assignments, and whether this differs from the general results. If this is the case, the best she can do is simply to take the third-personal data as it is. Further, as she cannot imaginatively access the outcomes of what it would be like to have a child – particularly the exact child she mothers, with its unique personality, physical attributes and health, and the particular attachment relation she will stand in to that child – she cannot inform her preference on this basis. So if she is to rely on third-personal evidence to make her decision rationally, she decides not to have a child. The problem is now that, though her decision is rational, it is inauthentic. It seems as though Sally has given up commitments she takes to be fundamental to her sense of self and her life plan. Paul’s challenge is that if we amend our notion of rationality to allow us to choose rationally in cases of transformative decision-making, the standard of rationality would be such that it would do offence to the authenticity of our choices.
But Paul does not think the story ends here. Perhaps we could reframe the decision problem in terms of deciding whether we want to discover what it would be like to have a child and who we would become after the experience. If one has had prior experiences of the relevant sort (about making major life-changing decisions), one could in principle attach a subjective value to discovery itself. If this is possible, Sally can rationally choose to have a child on the basis of discovery. The problem with this proposed solution is that to choose on the basis of discovering what an outcome will be like, including how it will change us, is to admit that there is nothing of ourself at present that we would not allow to be up for grabs. Any of our core preferences may change as the result of the transformative experience and they may change in ways that we consider to be at odds with our sense of identity. So again, there appears to be an affront to authenticity. Of course, the treatment here is simplified and does not do full justice to the nuance of Paul’s arguments.
Nonetheless, I ask: is this a legitimate challenge to decision theory? I think not. Firstly, it must be noted that Paul’s conception of decision theory is rather non-standard. The account is mentalist (it treats beliefs and values as psychologically real entities that an agent can look to when making decisions) and it is focused on the phenomenal (what it is like to experience an outcome). Traditional accounts of decision theory (see Von Neumann and Morgenstern or Savage) are ‘preference-first’ in the sense that it is a consistent set of preferences which determine whether or not an agent is rational. Consistency constraints are given in the form of intuitive axioms. For example, in order for me to be rational, it cannot be the case that my preferences are intransitive: I prefer A to B and B to C, but C to A. If my preference meet such axiomatic standards, then they are rational, regardless of what it is that I prefer. That is to say, justified beliefs and justified values do not combine to give us rational preferences, but rather, from rational preferences, beliefs and values can be described which mathematically capture an agent’s preference ordering. To explain this adequately would require some mathematical set up, which I will not do here. It suffices to say that on the more standard framework, Sally’s decision to have a child is rational insofar as it is consistent accordance with all of her other preferences.
In fact, the mentalist interpretation of decision theory is difficult to justify. What ordinarily grounds the idea that we ought to maximise expected utility is that we have more intuitive constraints on preference orderings, and given this, it follows that we maximise expected utility. This is known as a representation theorem. Without the axiomatic constraints on preferences, what tells us that maximising expected utility is the rational way to act? Furthermore, without a principle that tells us how to connect our psychologically real beliefs and utilities to the mathematical beliefs and utilities which characterise an agent’s preference ordering, Paul would not be able to claim Sally is irrational. That is to say, as long as Sally’s set of preferences is consistent with itself, she can be described by beliefs and utilities which make it such that she is maximising expected utility. It need not be the beliefs and utilities to which she actively looks when making decisions (unless we accept Paul’s very non-standard account of decision theory).
Even if we are to grant the normative standard is justified by some other means on the mentalist account, problems arise in this focus on the phenomenal value of experience and how one comes to know it. It appears to me that Paul holds preferences to an impossible standard. She holds that a preference is rational and authentic if the subjective values informing it are based upon imaginative forecasting of the outcomes. This imaginative forecasting is inaccessible to an agent in transformative situations due to epistemic limitations and personal transformations. Therefore, they cannot meet the standard of Paul-rationality. If these choices are to be made rationally in some way other than via imaginative forecasting, for example by use of third-personal evidence, they are doomed to be Paul-inauthentic. However, if the rationality and authenticity of a preference requires that that they are based on information that is by definition unavailable to the agent prior to making the decision, then this appears too strong a notion of informed preference. Further, this requirement would entail that Sally’s initial preference to have a child is equally inauthentic, as it cannot have been reached by such a process. If Sally’s initial preference is inauthentic, decision theory is not demanding that we change our preferences in a way that renders them inauthentic, and Paul’s argument does not get off the ground.
Whilst I do not think that Paul’s argument is sound, this is not to say transformative experiences do not raise questions for decision theory. Paul is right to emphasise the uncertainty and inaccessibility we face when deciding to undergo lived experiences that may fundamentally change us in ways unknown. Her discussion brings to bear the difficulty that is faced when we apply seemingly elegant and intuitive normative standards to muddled, real life decisions. It brings to bear that such decisions affect different interpretations of decision theory in different ways. It also brings to bear that standard decision theory pays little attention to issues of authenticity. Further work for decision theorists should be focused on specifying how significantly bounded agents are to make normatively rational choices in real life, particularly when the stakes are so high that they may threaten the agent’s sense of self.
Saira Khan is a doctoral student in the department of Logic & Philosophy of Science at University of California, Irvine.