Akrasia and the divided soul

‘Akrasia’, a Greek term used to indicate an agent acting against his own better judgment, has been widely explored in philosophical literature as an example of human irrationality. In what can be considered the beginning of this exploration, the Socratic dialogue Protagoras, Plato suggests that akrasia is impossible, but later ancient and contemporary literature continues to discuss akrasia as a prevalent phenomenon that should be better understood. Donald Davidson, in a paper entitled “How Is Weakness of the Will Possible?,” concludes that akrasia is indeed possible, but should ultimately be considered irrational. Using Plato’s conception of the divided soul in the Republic as a basis for my own interpretation, I suggest that in order to reach a proper understanding of akrasia we should take into account the possibility of mental fragmentation, and I intend to demonstrate how, within the fragmented mind, akratic action is impossible. In this paper I will clarify the particular phenomenon which I understand ‘akrasia’ to indicate, discuss Plato’s and Davidson’s contributions to the dialogue concerning this particular phenomenon, propose a schema for fragmentation that I believe appropriately accounts for human cognition, demonstrate why this schema renders akrasia impossible, and present reasons why fragmentation, rather than akrasia, is a more appropriate explanation of the behavior of individuals that we are tempted to describe as weak-willed.

The root of akrasia is the Greek word for power or strength, and the word itself suggests a certain lack of these qualities. For this reason, akrasia is usually equated with terms like “weakness of will” or incontinence, but because these phrases might carry connotations that may confuse the phenomenon which I intend to discuss, a clarification is necessary. In both Plato and Davidson, an akratic agent is one who holds the belief that a certain action is worse than a possible alternative and yet performs the action nonetheless. In the context of this discussion, akrasia exists outside the realm of any objective standards of goodness, and also does not concern an agent’s failure to apply an outward or universal standard of morality to his own behavior – akrasia should only be located in an agent’s performance of an action which he believes is wrong or worse for himself to perform, regardless of his beliefs about what others or ‘people in general’ ought to do. Returning to weakness of will and incontinence, these terms are appropriate to the extent that they imply a certain impotence of the will, but they should not, in the context of this paper, be considered synonymous with akrasia if they suggest a conception of weakness that is broader than an agent’s acting against his own better judgement — this will become more significant as my argument progresses.

With this understanding of akrasia as a foundation, it is easy to make sense of certain earlier Platonic claims, in the Socratic dialogue with Protagoras, that akrasia is impossible. Towards the end of the dialogue, Socrates, speaking with Protagoras, makes two intuitive assumptions: (1) for the individual, good is ultimately associated with pleasure, whether immediate or at some point in the future, and the same relationship holds with evil and pain, and (2) examples of seemingly akratic behavior are marked by an individual being “overcome with pleasure.” He then makes the following assertion: “The argument is absurd which affirms that a man … knowingly refuses to do what is good because he is overcome at the moment by pleasure.” To explain this, Socrates draws out the first association mentioned earlier, and continues:

Let us go on to say that a man does evil knowing that he does evil. But some one will ask, Why? Because he is overcome, is the first answer. And by what is he overcome? the enquirer will proceed to ask. And we shall not be able to reply ‘By pleasure,’ for the name of pleasure has been exchanged for that of good. In our answer, then, we shall only say that he is overcome. ‘By what?’ he will reiterate. By the good, we shall have to reply.

Plato’s conviction is that “nobody errs willingly” — he is convinced that man is only capable of performing actions he judges good. However, Plato demonstrates he is not unaware of an important distinction, included in my own definition of akrasia, between a greater good and a lesser one:

In the choice of … the greater and the less, and the nearer and remoter, must not this measuring be a consideration of their excess and defect and equality in relation to each other? … And this, as possessing measure, must undeniably also be an art and science? … Men err in their choice of pleasures and pains; that is, in their choice of good and evil, from defect of knowledge … that particular knowledge which is called measuring. And you are also aware that the erring act which is done without knowledge is done in ignorance. This, therefore, is the meaning of being overcome by pleasure — ignorance.

An agent’s failure to act towards the greatest possible good, Plato concludes, is indicative not of weakness but of ignorance, and it appears that since our definition of akrasia concerns an individual acting on or against his own judgement, any ignorance of the science of measurement cannot render an agent akratic, because ignorance suggests a certain lack of judgement in the first place.

Plato, via Socrates, finds it impossible that an agent can willingly act against his own better judgement, and he concludes that akrasia is a non-phenomenon. Christopher Bobonich affirms this understanding of Plato’s conclusion, suggesting “Socrates would have denied the possibility of what we call ‘weakness of will,’ since the only ‘will’ or rational desire there is is directed at something good, and never at something bad, and it is impossible not to desire the good” (xvii). However, while Plato’s conclusion initially seems both reasonable and attractive, the broader discussion of akrasia certainly does not end here.

Perhaps the most obvious source of dissatisfaction with the Protagoras conclusion can be located in its assertion about ignorance of measurement — while this ignorance is certainly possible, it seems that there are indeed cases where an agent is adept at the “science of measurement” but nevertheless chooses the good he judges lesser over the good he judges greater. One common example (that was fittingly not available to Plato — although there were certainly congruent cases) is the 21st century smoker who desires good health. Today’s smokers are typically knowledgeable, to a significant degree, that smoking greatly increases one’s risk for life-threatening disease. In addition, a significant percentage of these knowledgeable smokers would probably acknowledge that the pleasure they receive from smoking is measurably less than the pleasure of not having lung cancer, and yet millions of individuals who fit this description continue to smoke. Example like these render Plato’s measurement argument less than satisfactory. In light of this dissatisfaction, it will be helpful to consider one of the more influential modern explorations of akrasia, Davidson’s “How is Weakness of the Will Possible?”

Davidson is clearly aware of Plato’s assertions about akrasia, and in a later paper entitled “Paradoxes of Irrationality” he defines the Plato Principle as the “extreme” view that “no intentional action can be internally irrational” (175). Against Plato’s conclusion, however, Davidson proposes an understanding which does allow akratic agency, which he refers to as incontinence. I believe Davidson’s conception of incontinence is consistent with my definition of akrasia, and so I will use the two interchangeably in my recreation of his argument. Davidson proposes the following characterization of incontinence:

In doing x an agent acts incontinently if and only if: (a) the agent does x intentionally; (b) the agent believes there is an alternative action y open to him; and (c) the agent judges that, all things considered, it would be better to do y than to do x. (22)

He then proposes two principles which, to a certain extent, mirror Plato’s convictions in Protagoras:

P1. If an agent wants to do x more than he wants to do y and he believes himself free to do either x or y, then he will intentionally do x if he does either x or y intentionally.

P2. If an agent judges that it would be better to do x than to do y, then he wants to do x more than he wants to do y. (23)

Finally, acknowledging what seems like a preponderance of evidence for incontinence, Davidson asserts a final principle:

P3. There are incontinent actions.

It appears initially that the set P1-P3 is contradictory, but, while Davidson explores possible reasons to doubt each principle, he ultimately proposes that a contradiction does not exist. His proposal rests on a particular element of his characterization of incontinence, namely that the incontinent agent ultimately judges “all things considered,” and on a distinction between such conditional, or prima facie, judgements and unconditional judgments. He makes a keen observation about the phenomenon of moral conflict – whenever such a conflict exists, certain judgements must necessarily be made prima facie, otherwise even a merely conflicted agent would be guilty of holding contradictory beliefs. For example, a man who is considering cheating on his wife might consider that it would be better not to do so, in light of the fact that it would make her unhappy, but also that it would be better to do so, in light of the fact that it would be pleasurable. Because each of these judgements is made prima facie, in light of a certain set of evidence, an agent who undergoes moral deliberation can still be considered rational, because pf(x > y, a) does not contradict pf(y > x, b). However, moral principles, or the final judgements made in the resolution of conflicts, are understood by Davidson as unconditional judgments, which take the form of simply x > y or y > x, and do not contract either of the conditional judgments described above. Davidson suggests that while the judgment made by the agent in P2 is unconditional (it would be better to do x than to do y), judgements made by an incontinent agent are made prima facie (all things considered it would be better to do x than to do y), and concludes, in light of this distinction, that P3 does not logically contradict P1 and P2.

While Davidson’s argument is logically sound, it is importance to recognize, as Davidson himself does, that an essential component of the reasoning process is ignored by his incontinent agent:

How is it possible for a man to judge that a is better than b on the grounds that r, and yet not judge that a is better than b, when r is the sum of all that seems relevant to him? When we say that r contains all that seems relevant to the agent, don’t we just mean that nothing has been omitted that influences his judgement that a is better than b? (40)

Davidson ultimately concludes that while akrasia is not logically contradictory, and therefore possible, is it nevertheless irrational. He suggests that it is necessary that a rational agent “give … credence to the hypothesis supported by all available relevant evidence,” an understanding which motivates what he calls the principle of continence: “perform the action judged best on the basis of all available relevant reasons” (41). The consequence of Davidson’s conclusion, however, is that when an agent acts incontinently, he necessarily does so without reason — there is no reason why the action he takes prevails over the one he deems appropriate (see footnote 25). I find this conclusion less than satisfying. It seems clear to Plato that seemingly akratic behavior is caused by some cognitive phenomenon, which he describes as being “overcome by pleasure,” and even if Plato’s own conclusion is in Protagoras is unsatisfactory, it better approaches the question that I believe most people are actually interested in — not “how is akrasia possible?” but “why is a seemingly irrational phenomenon so common among persons we like to think of as rational?” Motivated by the latter question, I intend to provide a response to both.

The strategy I would like to pursue is one that assumes the reality of mental fragmentation, which implies the existence of two or more fragments within the mind that are, to a certain extent, independent of one another. A conception of fragmentation can be found in both Davidson’s “Paradoxes of Irrationality” and Plato’s Republic, each of which were written after their work which I have cited as explicitly addressing akrasia, but neither author fully pursues the conclusion which I believe will provide proper insight into problem of seemingly akratic behavior. I will, however, provide a summary of Plato’s divided soul in the Republic because it offers an appropriate foundation for my own conception of fragmentation.

In Republic Book IV, Plato proposes a tripartite view of the soul which consists of three entities whose names are translated as logic, appetite and spirit. Christopher Shields suggests that these parts can represent homunculi — distinct, autonomous (or semi-autonomous) ‘persons’ within a single mind. He offers the following characterization of the parts of Plato’s divided soul:

• (i) each part has pleasures and desires peculiar to it

• (ii) each part has its own wants and wishes

• (iii) parts discourse with one another, each cajoling and persuading the others of the superiority of its point of view

• (iv) each part can reason, or at least engage in rudimentary cognitive activities (Bobonich 80)

While I do not intend to explain akrasia using Plato’s tripartite view, I find it significant that one can elaborate a conception of fragmentation that is as old as the akrasia debate itself, and Shields’ characterization, which is derived from the Republic, is quite similar to my own. Shields himself is not sympathetic to a homuncular interpretation of Plato, nor to the use of homunculi in understanding akrasia, but because his conception of akrasia is broader than both mine and Davidson’s, and because his task is ultimately to clarify Plato, not akrasia, I do not find it problematic to use his homuncular interpretation of Plato as a means to my own ends.

Before delineating how a fragmented mind can make sense of akrasia, it will be helpful to explain why I find parts (i), (ii), and (iv) of Shields’ homuncular characterization appropriate. In addition, because I do not intend to appeal to Plato’s tripartite view, it will also help to introduce what I find to be a more appropriate division. Suppose there are two fragments, one of which has more holistic access to experiential evidence and, because of this, broader inductive reasoning abilities than the other. One might understand these fragments in a Freudian sense as conscious and unconscious, in a somewhat evolutionary sense as rational and instinctive, or even as similar to Plato’s logical and appetitive fragments, but because such labels have potential to confuse the goal of my distinction, I will refer to them as Fragment A (more holistic) and Fragment B (less holistic).

I am partial to Hume’s understanding that sensory impressions are the fundamental entities which make up the mind, and therefore I find it easy to suggest that any distinct psychological system has access to a certain set of experiential evidence, and that access to this evidence constitutes sufficient reason to suppose that such a system is capable of forming its own propositional attitudes. If this can be accepted, Shields’ understanding that each part of the soul is capable of having a distinct set of desires and wishes is appropriate, and it should also follow that each part is capable of a distinct set of beliefs as well. However, Shields’ final characterization, that each part is capable of reason or at least rudimentary cognitive processing, may be more difficult to accept as applicable to the fragments I have suggested.

I suspect that some readers will find this final characterization problematic when it is applied to Fragment B, which I have referred to as unconscious or instinctual. I am not convinced that it is. It seems appropriate to suggest that any distinct psychological system which can translate a collection of sensory impressions into action is at least capable of forming propositional attitudes and processing them in a manner than is logically consistent. When my dog, whose psychology might be described as instinctual, experiences the sound of his food falling into his bowl, he consistently comes running to eat. My dog’s consistent action in this regard should be attributed to some type of cognitive association between the sound and the presence of food. I see no reason why this association should not be interpreted as a belief that there is food in his bowl, and this belief, together with a desire for food, can be understood to cause a logically appropriate action. Whether or not my dog has the ability to deliberate and judge that some other course of action would be better is inconsequential — his attitudes are founded on evidence and his action is rational given his attitudes. If action is difficult to attribute to a psychological system, as a physical rather than a cognitive phenomenon, one can still attribute rational intention to the psychological system in question. It should be clear that if psychological system within my dog can perform certain cognitive functions which ultimately result in logically consistent behavior, any distinct psychological system within the human mind should be able to do the same.

If I have established that both Fragment A and Fragment B can be attributed a unique set of beliefs and desires, as well as the ability to rationally translate these into intentions and potentially actions, I can now direct attention back to akrasia, namely Davidson’s seemingly contradictory set P1-P3. In P1 and P2, Davidson describes an agent that has the belief that x is better than y, the desire to do x more than y, and the intention to do x instead of y. If mental fragments have their own beliefs, desires and intentions, Davidson’s agent need not necessarily indicate an individual — it could instead indicate one of an individual’s mental fragments, which is essentially an agent in itself.

A new picture of the apparently akratic individual is now available. Within the framework set forth by Davidson, an individual can make the prima facie judgment, x is better than y all things considered, in Fragment A, take the rational step, under Davidson’s principle of continence, to an unconditional judgment, x is better than y, again in Fragment A, and nevertheless intentionally do y based on an unconditional judgment in Fragment B. An agent has rationally and intentionally acted based on a coherent set of beliefs and desires which can be attributed to him, and yet we might be tempted to label the individual akratic because he appears to simultaneously believe something contradictory. Based on Davidson’s own definition of akrasia, however, this label is not appropriate — no unique agent has acted against their own better judgement. If mental fragmentation can explain that what appears to be akrasia is actually a different mental phenomenon, akrasia can be considered impossible, P3 can be discarded, and we can approach a more explanatory conclusion than one that suggests the broader phenomenon of human weakness is best understood as simply irrational.

I find it appropriate to illustrate how Fragments A and B, as distinguished earlier, can make sense of the example we were concerned with earlier, the 21st century smoker. Fragment A, with its more holistic set of evidence and broader inductive reasoning abilities, can make the judgment that, based on scientific, peer-review studies about the effects of smoking, it would be better to quit smoking than continue. Fragment A can be considered fully capable of Plato’s “science of measurement.” Fragment B, on the other hand, is limited to a more immediate cognitive relationship between smoking and pleasure. Both fragments both ultimately desire pleasure, but Fragment B’s rational, all things considered, and therefore unconditional judgement is that it would be best to continue smoking, and the 21st century smoker can be considered to act on a rational intention formed in Fragment B.

If this less than convincing, imagine someone we might refer to as ‘strong-willed.’ We might attribute beliefs to this individual based on what he says he believes, and assume he has a strong will because his actions correspond with his professed beliefs. When this individual acts in a manner that is consistent with his professed beliefs, we suggest that his action was intentional, and that his intention was derived from his beliefs. I think the typical observer is especially prone to such an interpretation when these beliefs are oriented away from immediate pleasure – when it might be considered harder to conform to them. However, in cases like Plato’s, when an individual is “overcome by pleasure,” I see no difference in the relevant psychology. As I have demonstrated, the pleasure-oriented fragment is capable of having its own beliefs, one of which could be understood to take the form of the maxim ‘immediate pleasure is always the greatest good,’ and actions which seem to follow this belief can easily be understood as motivated by rational intention. The fact that this fragment’s beliefs about the greatest good happen to always correspond with its immediate desires should have no bearing on the rational nature and intentionality of actions which are performed by an individual when this fragment’s line of reasoning is implemented. If the schema of fragmentation I have suggested is reasonable and can account for consistent paradigms of human behavior, I believe it is more appropriate to attribute behavior to a coherent agent within the individual than to suggest the individual in question is simply ‘overcome’ or even coerced by a form of pleasure or desire that, in this sense, seems almost external to the individual himself.

During his discussion of the tripartite division of the soul, Plato suggests that within the “just man,” the logical division, which can be considered equivalent to Fragment A, is always triumphant. One might be tempted to consider Plato’s just man an appropriate example of the strong-willed individual. It should be recognized, however, that within the just man, a certain weakness of will is still apparent — only now Fragment B, or Plato’s appetitive division, demonstrates the weakness or incontinence. Plato himself alludes to this paradox in the Republic:

There is something ridiculous in the expression ‘master of himself’; for the master is also the servant and the servant the master; and in all these modes of speaking the same person is denoted.

Nevertheless, he persists in asserting that just or strong individuals should rule themselves with their logical division:

The meaning is, I believe, that in the human soul there is a better and also a worse principle; and when the better has the worse under control, then a man is said to be master of himself; and this is a term of praise: but when, owing to evil education or association, the better principle, which is also the smaller, is overwhelmed by the greater mass of the worse — in this case he is blamed and is called the slave of self and unprincipled.

Plato’s conclusion seems to suggest that the just man’s logical part will always be implementational (action will always proceed from this division’s intention), or that his logical part will always successfully convince his appetitive part to agree with the former’s conclusions. However, within this characterization, while a distinct agent never demonstrates akrasia by my narrow definition of the term, individuals whose logical division fails to be authoritative can still be considered somewhat weak-willed.

While Plato’s convictions are intuitive, I believe there is room to push back against his characterization. An understanding of evolutionary psychology and modern cognitive science suggests that there is a particular merit to the less holistic and more instinctual or unconscious elements of human behavior. Cognitively, a less holistic module is computationally more adept at responding to immediate danger. In addition, the potentially overwhelming desire to eat or drink, which might be located in primarily in Fragment B, is not something a healthy human should want to master completely, for example, at the hands of some rationalized ascetic philosophy. Within an individual convinced of such a philosophy, we might locate the belief that eating or drinking is wrong in Fragment A, but this individual’s ‘weakness’ in conforming to such a belief, which we can understand as the implementation of Fragment B, can be understood as beneficial. The desire for sex, which can be easily be located in Fragment B, is another good example – this desire, which many might understand as somewhat animalistic and perhaps even irrational, is responsible for the maintenance of our species.

I have already alluded to the argument which Plato makes against this line of reasoning. The Fragment B desires for food, drink, or sex, especially as I have characterized them, do not have to be understood as necessarily inconsistent with the measured desires we might attribute to Fragment A. However, when Fragment B desires these to an unhealthy extent, Plato would suggest it would be weak or unjust to continue to indulge oneself against the judgement of Fragment B. But if we can lend even a small degree of legitimacy to the function of Fragment B, it seems much less appropriate to refer to conflicted individuals as weak-willed, because when we do so we unwarrantedly locate the self in one fragment and not in both. To understand exactly why labels like akrasia and weakness of will are inappropriate if we have reason to understand the individual as the sum of its fragments, consider the following example. If we can attribute the belief ‘immediate pleasure is the greatest good’ to one of an individual’s fragments, whenever an individual acts contrary to this belief we might call him weak-willed, but in doing so we classify someone who would typically be considered strong-willed as exactly the opposite. In light of this circularity, the following should be clear: if we accept fragmentation, the debate about ‘acting against one’s better judgement’ is simply indicative of a misunderstanding of the divided soul.

I acknowledge that my reinterpretation of the broader problem of seemingly akratic behavior leaves us right where we started, grappling with the phenomenon of a certain impotence in directing our actions with what feels like our own rational will, or the holistic mental processes which occurs in Fragment A. I would suggest, however, that my appeal to fragmentation sheds light on a more satisfying outlook on the problem of weakness. Because each of their fragments can be considered motivated by both evidence and reason, individuals who struggle with weakness of will might consider being less concerned with strength and more concerned with unity.

Works Cited

Bobonich, Christopher, and Pierre Destrée. Akrasia in Greek Philosophy. 1st ed. Leiden: Brill, 2007. Print.

Davidson, Donald. “How Is Weakness of the Will Possible?”. Essays on Actions and Events (2001): 21-42. Web. 17 Apr. 2017.

Davidson, Donald. “Paradoxes of Irrationality”. Problems of Rationality (2004): 169-188. Web. 17 Apr. 2017.

Plato. Protagoras. Trans. B. Jowett. The Internet Classics Archive. Massachusetts Institute of Technology, 13 Sept. 2007. Web. 17 Apr. 2017.

Plato. Republic. Trans. B. Jowett. The Internet Classics Archive. Massachusetts Institute of Technology, 13 Sept. 2007. Web. 17 Apr. 2017.