The main motivation for NNs is that they modify offender behavior, thereby helping to rehabilitate offenders and reduce recidivism. These intended effects work by modifying subjects’ brain states and corresponding mental states. Since NNs are non-consensual, and do not engage with the subjects’ capacities for control over their mental lives, their effects on subjects’ mental states occur via bypassing. Although they share this feature, neurocorrectives are quite diverse in their methods, intended effects, and degrees of efficacy. What bypassing views have to say about specific sorts of NNs will depend on the particular features of the neurocorrective in question, and on the nature of the subject of the NN. My focus will be on what general claims we can make about the effects that NNs might have on subjects’ freedom.
Yet we can still narrow down the subject matter somewhat. As previously mentioned, it is plausible that we have some reasons not to implement NNs. The reasons in favor of implementing them, reasons which might outweigh the reasons against, concern the possibility of rehabilitation of offenders and reducing recidivism. Neurocorrectives which do not help in achieving these ends will likely not be good candidates for NNs, since there will not be reasons in favor of their implementation that plausibly outweigh the reasons against. Thus, such neurocorrectives will not be of great concern for our purposes here.
This section is divided into three parts. First, I consider questions relating to whether NNs might undermine an agent’s freedom for some actions. Second, I discuss how subjects whose freedom for some actions has been undermined may regain freedom for these sorts of actions. Finally, I consider the possibility that freedom comes in degrees, and what implications this has for our main question.
One important feature of NNs to keep in mind is that, as they are typically described, they are targeted. That is, they are intended to work via the production of a specific set of attitudes—say, those arising from increased empathy—or the elimination of a specific set of attitudes—say, the sexual urges meant to be eliminated by anti-libidinal neurocorrectives. An NN that is relatively targeted is unlikely to undermine freedom for all of a subject’s actions. The attitudes that were acquired or significantly changed via bypassing will not be relevant to all contexts of deliberation, and the subject will likely perform a variety of actions that do not issue from such states. Even if, for example, an anti-libidinal drug significantly modifies some of the subject’s values and desires, the subject may still freely make a decision about whether to pay a visit to his ailing mother. If NNs undermine freedom for actions, it will likely only be with respect to some actions.
One feature that bypassing views agree on is that, in order to undermine freedom for some action, the change that occurs via bypassing should be significant, where significance can be roughly understood as a measure of the breadth and depth of the change. The more significant the change that a particular NN produces, the more likely it is to meet any particular significance criterion. Consequently, the more significant the change produced, the more likely that the NN will undermine freedom for some actions. How likely this is, I suggest, will depend on the state’s policy, and its expectations.
First, the goal of rehabilitating offenders and reducing recidivism will be importantly related to the significance of the change. If the most that a particular NN does to any individual is produce changes that are not relevantly different from how “a momentary alteration in attention due to bad digestion might affect someone’s deliberation or subsequent decision, or a quick spike in blood sugar” (McKenna, 2017, p. 579), then it is doubtful that the NN will be of much help in rehabilitating offenders or in reducing recidivism. NNs intended to achieve these goals would likely need to be more significant, and thus risk undermining subjects’ freedom with respect to some actions.
Further, the significance of the change required to achieve these goals will also be related to the nature of the offenders that the NN is intended to change. Rehabilitating some subjects may not call for a very significant change in attitudes, whereas rehabilitating others may require a very significant change. Given that NNs are non-consensual and often perceived to be quite invasive, they are likely to be reserved for perpetrators of particularly serious offenses, and/or for repeat offenders. Suppose that the state restricted the implementation of NNs to such offenders.Footnote 25 Such a policy would seem to select for subjects whose rehabilitation calls for a fairly significant change and thus, for NNs that produce one.
These points apply to effects of NNs that help to achieve the goals of the state. However, as is sometimes the case with biological and neurological interventions, NNs may in some instances result in unintended side-effects. In relation to this, it is important to note that even if an effect is unintended, it can still be a change that occurs via bypassing. Consequently, such side-effects will need to be taken into account in the same way we would take the intended effects into account.Footnote 26
Side-effects resulting in changes via bypassing can widen the range of actions for which freedom is threatened. Suppose, for instance, that an empathy increasing drug has a side-effect resulting in the subject gaining a preference for comedy movies. This might then threaten her freedom for cases in which she is making a choice between two such movies, if the change is significant enough.
However, gauging the likelihood that an NN will undermine freedom in this way will be more difficult. As I suggested above, one of the important factors in selecting which neurocorrectives to administer as NNs will be the significance of the intended effect; in order to aid rehabilitation and the reduction of recidivism, the neurocorrective will likely need to produce significant intended effects.Footnote 27 This does not give us much guidance in relation to side-effects, since significant side-effects will not provide a reason in favor of implementing the NN; in fact, they would seem to provide a reason against. For similar reasons, we would not get much guidance on this from the point concerning subject-selection.
So far, I have considered some of the relevant factors in determining whether an NN will undermine offenders’ freedom for at least some actions, including the significance of both intended and unintended effects, and the nature of the state’s policy, including which sorts of offenders are selected as subjects. However, these are just general points; particular NNs will likely affect different people in different ways, and these effects are unlikely to be uniform.
In cases where NNs have an effect that meets the significance criterion, we seem to have a pro tanto reason to refrain from using them. This is not to suggest that this is the only reason to refrain from using an NN in a particular case, nor is it to suggest that this reason overrides all other reasons in favor of implementing it; as suggested earlier, there are many considerations to take into account. Yet one might object to the claim that we would have such a pro tanto reason. One might think that if an NN eliminates offenders’ freedom to commit serious offenses – say, to murder or rape – this does not give us any reason to refrain from implementing them, since this is not a valuable form of freedom. Thus, even if NNs undermine subjects’ freedom with respect to some actions, this does not count against their implementation.Footnote 28
In response, I wish to suggest that the issue here is not with offenders’ freedom to commit serious offenses, but rather with whether they freely perform other actions. The fact that an NN might eliminate, or perhaps merely mitigate, freedom with respect to these actions provides us with a pro tanto reason to refrain from their use, and this does not imply that the freedom to commit serious offenses is valuable.
To see why, consider the case of Paul again, the mediocre father whose evaluative scheme was radically modified and as a result, decides to take out a substantial loan to finance his daughter’s education. He does not freely decide to do so, and bypassing views can account for this. However, note that the claim here is not that he does not freely decide to take out the loan because he could not have done something else. Pat, the unmanipulated father, also has a Luther-style inability to do otherwise, yet this does not mean that he does not freely decide to take out the loan.
A similar point can be made concerning the claim that NNs undermine freedom for some actions. Consider a couple of cases. First, suppose that rather than implementing the neurointervention nonconsensually, the state implements the neurointervention only after discussions with the offender reveal that he wants to take it, and the offender gives valid consent to the intervention.Footnote 29 Second, suppose that, after much careful research, we discover a new pedagogical technique that makes it possible to rehabilitate some of these offenders in ways previously thought unimaginable. Further suppose that this technique robustly interacts with the agent’s capacities for control over his or her mental life.Footnote 30
In both of these cases, the subjects may undergo changes as significant as those experienced by the subject of an NN. Further, they may lack the freedom to commit serious offenses, at least to the same extent as a subject of an NN. Yet, although bypassing views do not imply that the subjects fail to act freely in these two cases, they may for the subject of an NN. This difference is not explained by a difference in terms of their freedom to commit serious offenses, since there is no such difference. Thus, it is not clear how the supposition that we have a pro tanto reason to refrain from using NNs, in virtue of the fact that they would result in some subjects not acting freely in some cases, implies that the freedom to commit serious offenses is of any value.
If an NN undermines freedom for some actions, then we would seem to have some reason not to implement it. Yet whether or not agents can regain freedom for these actions, and how easy it would be to do so, may either affect the strength of this reason, or provide us with further reasons against the implementation of NNs. Prima facie, at least, it would seem that the longer the period of time that an NN negatively affects freedom, the stronger the reason(s) we have against its implementation. As suggested above, there is no clear answer to the hard problem of regaining freedom. Until such an account is provided, we may need to settle for some uncertainty with regard to this question.
This is one place in which the targeted nature of NNs may make a difference. If an NN is fairly targeted, and does not result in changes that would significantly affect the agent’s evaluation of the changes that occurred via bypassing, then we may get something akin to one of the easy cases of regaining freedom. Suppose, for instance, that a subject of an empathy-increasing NN is now better at recognizing when her actions may harm other individuals. Further suppose that she already believed that she should avoid harm to others, though she did not always recognize that some of her behavior did harm to others. In such a case, the subject may end up identifying with, and endorsing, the increased empathy on the basis of previously held attitudes; and this might be a relatively quick process. This is a straightforward case of coming to be free with regard to actions issuing from these attitudes.
However, it is doubtful that all cases of NNs will have such happy results. Suppose that a subject of an empathy-increasing NN did not believe that one should put much weight on the suffering of others, or suppose that the subject of an aggression-lowering NN took pride in his aggressive impulses, thinking that they were appropriate. In such cases, and supposing that the NN eliminated the agents’ freedom for some actions, regaining freedom may be quite difficult. Such cases may be more like a version of the Judith case in which she does not want the strong desire to punch her friend in the face that was implanted in her.
Above, I suggested that in a case like this one, in which the agent determines that she wants to change or eliminate the attitude on the basis of previously held attitudes, then insofar as she retains this disposition towards the implanted or changed attitude, actions issuing from it would not be free if a) she has not had the opportunity to modify or eliminate it, and b) she has not had the opportunity to learn how to reliably resist it in the contexts like the current one. If the NN is targeted enough to make this an easy case of regaining freedom, and say, it is not difficult for the agent to modify the attitude, or learn to resist them reliably, then regaining freedom for actions that issue from these attitudes may again be somewhat straightforward.
Yet, here again we encounter an important tension between the goals of implementing NNs, and what would need to happen in order for the agent to act freely from such attitudes. Insofar as the state’s goal in implementing NNs is to rehabilitate offenders and reduce recidivism, they will have an interest in making it difficult for subjects of NNs to have either of the opportunities just mentioned. If doing so were easy, then it would be easy for offenders to significantly thwart the state’s goals, and would likely lead to a much less effective policy of NNs. Assuming that the state would want to prevent this, they would have some reason to make it difficult to have either of these opportunities. Perhaps this could be done by a single implementation of a powerful, and perhaps long-lasting, neurocorrective; or perhaps it could simply be done by a continual administration of the NN. For instance, if the NN is a drug, it may be administered periodically, and a re-administration may undermine much of the work done by an offender, or perhaps rejuvenate the NN’s effect.
It is important to note, however, that such a tension need not always be present, and this is because the subject’s disposition towards the changed attitudes can change over time. Thus, even if a subject of an NN is initially opposed to the change, and assuming that it was significant enough to eliminate freedom with respect to actions resulting from it, she may still come to be free with respect to actions issuing from the implanted attitudes without having either the opportunity to modify or eliminate the attitudes, nor the opportunity to learn to reliably resist it in similar contexts of practical deliberation. This can happen, for example, if the subject ends up identifying with or endorsing the attitudes in a way that is not wholly, or perhaps significantly, guided by further implanted attitudes. That is, she may re-evaluate the attitudes and come to be free with respect actions issuing from them if the case changes into one like the first sort of easy case discussed above. Supposing that this is possible, the state may then have reason to help the subject regain freedom in this way. Suppose, for instance, that there is a course of therapy that engages with the subject’s capacities for control over her mental life, and can help her come to identify with and endorse the implanted attitudes. Offering this therapy to the subject can then help her regain freedom with respect to the relevant actions in a way that is consistent with the state’s goals of rehabilitating offenders and reducing recidivism.Footnote 31
Degrees of freedom
Finally, one further consideration to take into account is the possibility of degrees of freedom. One might think that agents can be more or less free with respect to some action or other, and that the degree to which one is free is a function of the extent to which one exemplifies freedom-relevant features; e.g., how much control an agent has over some action, or how reasons-responsive the agent was at the time of deliberation and/or action.Footnote 32 Bypassing views, intended to explain why agents like Paul do not act freely, offer conditions on when bypassing can fully undermine freedom. However, the features picked out by bypassing views can come in degrees as well. For instance, the change effected via bypassing can be more or less significant, and one might have more or less of an opportunity to evaluate and modify attitudes acquired via bypassing. This has important implications.
First, this suggests that an NN can negatively impact freedom even if it does not fully undermine it with respect to some actions, in virtue of merely reducing freedom with respect to some actions. In such cases, we may still have a reason against the implementation of NNs. Moreover, the threshold that NNs need to meet in order to mitigate freedom will presumably be lower than the threshold they would need to meet in order to eliminate it; changes occurring via bypassing that are not significant enough to fully undermine freedom may still be significant enough to reduce it. Thus, it is even more likely that NNs will have a negative impact on freedom.
Second, recall that some have argued that NNs, or enhancements more generally, might increase an agent’s freedom in virtue of improving freedom-relevant features. The features that would be improved according to these arguments are synchronic features; e.g., reasons-responsiveness, or control at the time of action (or shortly before). These are the sorts of features that Pat and Paul can have to the same extent, even though Pat freely decides and Paul does not. This positive effect on synchronic freedom-relevant features is consistent with the negative effect on the diachronic features that bypassing views focus on. What should we say about these cases?
Since bypassing views offer a necessary condition on free action, failing to meet it with respect to some action implies that the agent does not act freely, regardless of whether the NN improved some synchronic freedom-relevant features. Suppose, for instance, that all four conjuncts of NFMR, the most extreme of the bypassing conditions, are true of a subject’s action. In such a case, the subject would not act freely; regardless of whether the NN improved the subject’s synchronic freedom-relevant conditions.Footnote 33
However, a different sort of case will be much more difficult to assess. Suppose that, although an NN does not fully undermine an agent’s freedom with respect to some action, it does have significant effects, and would typically be taken to mitigate freedom with respect to some actions. But, suppose also that the NN results in an improved capacity to recognize and respond to reasons. In such a case, determining the NN’s overall effect on the agent’s freedom with respect to some actions will be a difficult matter since presumably, the extent to which an action is free is at least partly a function of both the agent’s synchronic freedom-relevant features, as well as the features identified in bypassing views.
This is further complicated by the fact that a necessary condition on free action, as is proposed by bypassing theorists, does not give us a full account of free action; importantly, it does not clearly commit one to claims about which synchronic features are freedom-relevant features. For instance, theorists might disagree on whether, in order to act freely, one merely needs responsiveness to reasons, or whether one needs, further, responsiveness to moral reasons,Footnote 34 or, they might disagree on whether one also needs the ability to do otherwise at, or shortly before, the time of action.Footnote 35 Thus, even among those who adopt bypassing views, there is room for disagreement about when, precisely, an NN would improve synchronic freedom-relevant features.