By Paul Almond, 9 February 2006

Introduction

There is controversy about the extent to which we should hold people "responsible" for actions. In some situations we say that a person is "responsible" for an action, but in others, despite the act being committed, that he/she is not "responsible". What is this "responsibility"? If we can say when people have it and when they do not then we would be answering this question.

Common ideas about assigning responsibility are too vague or inconsistent to be philosophically useful. This article will suggest an approach that is consistent and could be formalized.

This article is not intended to resolve the issue of responsibility entirely, but merely to propose a very basic idea.

It will be obvious from my previous articles that I am interested in the prospect of machine intelligence. We may want to consider the idea of responsibility in such an area, or in relation to some other hypothetical form of non-human intelligence, and the idea proposed here would easily extend into such areas. For simplicity, in this article I will be referring to the system doing the thinking as a "brain" and, mostly, presuming that we are considering human responsibility.

Examples

How do we assign "responsibility" to people? We could say that if someone does something then he/she is "responsible". A few people may say that they have such a view, but most people would have problems with taking it in all situations. How would you feel about the extent to which we should hold someone responsible in these scenarios?

• A person who steals, simply saying that he/she can have other people's property if he/she wants. Beyond that we have no detailed idea of why he/she does it: no explanation presents itself, though we may have some ideas about his/her upbringing, events in his/her life, etc - basically the usual sort of story that you read about a defence lawyer using to try to get a lenient sentence - however, many people who had similar sorts of experiences do not steal.
• A person who routinely assaults other people. A psychologist says that he/she is psychopathic.
• A person with a known neurological disorder who assaults someone after stopping taking medication that controlled his/her symptoms.
• A person with a known neurological condition who assaults someone after his/her flatmate maliciously swapped the medication for something that appeared similar but was ineffectual at controlling the condition.
• A person who loses his temper and kills someone. He/she has a genetic disorder known as "Temperlossitus". 80% of the people who have it try to kill someone.
• A person with an extreme fever who is hallucinating and hits someone who tries to help.
• A person with a microchip involuntarily implanted into his/her brain, where the spinal cord joins it, able to override what he wants to do by sending its own signals down his/her spinal cord. The person finds him/herself aiming a gun at someone. He/she tries frantically to put the gun down, but the microchip has other ideas. He/she watches horrified as his/her own body refuses to obey him/her and his/her finger pulls the trigger.
• A person with a microchip implanted in his/her brain that makes him/her irrationally hate people whom he/she then assaults.
• A person who has had a drug put into his/her drink in a bar that makes him/her suggestible. He/she is told to commit violence and does so, later feeling remorse at what he/she did while under the influence of the drug.
• A person who has been indoctrinated into a religion and thinks that a particular group of people are representatives of the devil and need to be killed.

Maybe not all of these situations are plausible: they are not meant to be. The point is that most of us would accept that varying differing levels of responsibility apply in these situations.

The Lazy Fallacy of Free Will

A common belief is that someone is responsible for his/her actions if they are due to "free will", but less responsible if something beyond his/her control intervenes. This does not really say anything. What does "free will" mean? What does "beyond his/her control" mean?

If someone's brain were being disrupted by some neurological condition, a drug that he/she had been tricked into taking or an implanted device we may say that he/she was being influenced by something that was beyond his/her control, and most people would probably have the following view:

1. There is "free will" - that part of a person which decides on courses of action.
2. Physical events can occur to detract from a person's free will - some actions could be caused by these physical events rather than free will.

Many philosophers view "free will" as an incoherent concept. I certainly view it as such - at least in the sense that the word is most commonly used. I will not be going too far into that here, though. The implicit assumption here though is that what is called "free will" is somehow different from all the physical events that can compromise it. Brains, however, are physical systems. This suggests that all human actions have physical causes. How could we ever differentiate between those physical events that are under the control of a person and those that are not when the person's volition itself - what many would call "free will" - is actually the result of physical events? This point is made by Julian Baggini in What's It All About? Philosophy and the Meaning of Life [1]. Some people think that the non-determinism widely believed to be a feature of quantum mechanics provides a loophole here, but it does not: Baggini observes that such non-determinism is only at the subatomic level and has nothing to do with things on a human scale, that it does not allow for free will because a free choice is not a random choice and that quantum mechanics itself is incompletely understood now [2]. I would state it this way:

The claim of free will is essentially that there is a "you" who is not driven by physical events beyond your control. Even if the universe is throwing out random numbers it is those random numbers that control you. You do not control the random generator.

This issue is relevant in court cases. Some legal systems will view a defendant as having "diminished responsibility" - as being less responsible - if he/she is declared mentally ill or compromised in various ways [3]. The idea is that the mental illness detracts from his/her free will - something is wrong with his/her brain. A defendant who is unfortunate enough to lack a mental illness, on the other hand, would be held responsible.

Why do we tend to make such decisions? Many people would say that "something is wrong with the brain" of someone who is accepted as having diminished responsibility, but what does this mean? Presumably it means that some physical characteristic of the brain, labelled as a mental disorder, contributed to the action, but what about a person who is not accepted as having diminished responsibility? His/her actions were also caused by physical events in the brain. As he/she clearly has a brain that made him/her perform the action, and as the action is unacceptable, why do we not say that something is wrong with that brain? I am not saying that we necessarily should, but we do need to have some better grasp of this sort of issue. We cannot simply say that a person is less responsible when his/her behaviour is affected by a neurological condition: thinking itself is a neurological condition.

Some readers may object and say that the person who is mentally ill is made to perform various actions by the condition, whereas a person who is responsible is able to change the physical nature of his/her brain at any time. The problem with this is that a person is not independent of his/her brain. Any decision that a person makes to change the nature of his/her brain is being made by the brain itself, which means that it was in the nature of the brain to do it anyway. This is a point raised by Hofstadter [4]. I also discussed this issue in an earlier article [5].

Because none of us can stand outside of our physical nature, each of us is condemned to do whatever it is in that nature to do. We can never do otherwise, for if we chose to do something else that choice must have been in our nature at the time that we made it. At any time the human brain is in a physical state. It then moves on to another physical state, and another, and so on. Each time the next physical state that it assumes is controlled by its current physical state and whatever is happening to it in the world. None of this is controlled by "you" because this system is you.

The fallacy that somehow we can escape from this system is a common one. I have frequently heard people saying, "Yes, but you do not understand. We might be moving along the rails like trains, but we can always choose to move on to different rails." It is not really like this. A person may think that any time he may choose to change course, but the exercise of free will to change that course is only a perception rather than reality. From a physical point of view the system was on the real set of rails all along and when it "chose" to "change course" it was simply following rails that made it do as such

Given that all human actions are governed by what we physically are, when should we hold a person accountable for what he/she physically is and when should we hold a person less accountable?

Consistency with Intuition

"Responsibility" is probably an artificial thing. Beyond humans it is doubtful that the rest of the inanimate matter in the universe around us, or the laws of physics, really care too much about what humans do or how responsible they are.

Whatever standard is proposed for measuring responsibility should be consistent with what we generally understand "responsibility" to mean.

The solution should roughly match our general intuitive expectations. As an example, if it tells us that people against whom we feel total revulsion are not responsible for their crimes, but people with whom we tend to have sympathy are responsible, then we should call it into question. A solution may make us think about some of our intuitive beliefs but there should be some kind of correspondence. Otherwise it would simply be a system that nobody would take seriously anyway.

Complexity of Algorithmic Therapy on a Brain Description

Complexity of algorithmic therapy on a brain description is the term I give to the standard which I will propose. It would probably need a less cumbersome name if it caught on. The idea is simple:

First, imagine that we have a formalized description of the brain of the person. In previous articles I have discussed the well known concept of mind uploading. It is this sort of description I am talking about. Imagine that we had a machine which could somehow scan a human brain and produce a set of numbers encoding its description - which would include whatever state it currently is in. You do not need to think that this is practically feasible: we are simply using it as a philosophical device. The idea here is simply that we are going to consider the description of the person's brain.

The description of the brain is simply data: it is a sequence of numbers that can be processed by a computer program. If we knew how, we could write an algorithm which would take the sequence of numbers describing the brain in a particular condition, process it and produce a sequence of numbers describing a brain in a different condition.

The degree of responsibility which a person is presumed to have for undesirable behaviour is dependent on the minimum length of the algorithm needed to process a description of his/her brain and produce an alternative description of a "corrected" brain - a brain which people would regard as being "cured" of the problematic behavioural tendencies.

This means that we are imagining "correcting" the brain by using a computer program. If we had a digitised model of the person's brain and we used a computer program to effect a correction, how long would the computer program have to be? The greater the length of the program then the more responsible the person is held to be.

The idea here is really that we are interested in how "far away" the person is from behaving acceptably. If it turns out that the person's brain is almost behaving acceptably, but just needs a simple nudge in the right direction, the sort of thing that could be done by a slight modification of its description, then we may see the person as having little responsibility. If, on the other hand the brain is so far away from what we would consider acceptable that a very serious algorithmic intervention is needed, maybe requiring a computer program almost as complex as the brain itself, then we may be justified in thinking that it is less tenable that the person is somehow a victim of something else.

An Alternative Statement of the Idea

I have just stated the idea in terms of algorithm length. It is also possible to imagine it differently. An alternative standard would be to consider how many different algorithms are available to convert the problematic brain description into an acceptable one. The more such algorithms were available then the less responsible we would view the person as being. Of course, if algorithms can be of unlimited length then there are an infinite number of such algorithms available, so some kind of limit on length would need to apply, or some kind of comparison used as this limit tended to infinity.

Size of the Brain and Fairness

The algorithm needed to convert the descriptions will have some dependency on the size of the brain. A large, complex brain would need a very large algorithm to do almost anything with its description while serious changes could be made to a small brain with very little programming. This may seem to be unfair to anything with a really huge brain, or equivalent of a brain, that we may encounter. An example is the fictional Solid State Entity in David Zindell's science fiction novel Neverness [6] - an entity consisting of a network of moon-sized computing devices spanning many light years and with intelligence far beyond that of humans. Would this not mean that the Solid State Entity would always have utter responsibility for anything that it did, simply because it is so vast, and so intrinsically complex, that any algorithm which altered its description to produce any reasonable and controlled change in its behaviour would by necessity be much larger than any algorithm needed to perform the same sort of change in a human brain?

This need not really be a problem. We should be able to imagine some sort of normalisation whereby we take the length of the description into account. We may, for example, say that the degree of responsibility is dependent on the length of program needed to process the description divided by the number of bits in the brain description, or something similar to that.

Defining a "Corrected" Brain

This idea of responsibility relies on having an idea of what would constitute a description of a brain that has been "corrected" by an algorithm. This is not as simple as it may sound: after all, any unacceptable behaviour can be stopped very simply providing that you do not need to have any complex functioning afterwards.

Putting this rather crudely, a human brain that was behaving in a disturbing way could be "cured" by the expedient of putting a bullet into it. This is drastic, but it would almost certainly mean that the disturbing behaviour would cease. As a cure, however, it would probably not be considered viable: the side effects - cessation of any other behaviour as well - seem excessive. Some people may agree with capital punishment, but that is not the point here: unless we are going to us emotive language we would not pretend that destroying a brain somehow "corrects" it.

An intervention which is the algorithmic equivalent of a gunshot to the head could be made to correct any description of a brain: an algorithm could process the brain description and simply produce a sequence of 0s as the description of the "corrected" brain or randomly scramble neural pathways. It would then be argued that the new brain description is fine because it no longer exhibits the problematic behaviour. This would be correct, in a way, but not really because the problematic behaviour has been specifically corrected: it would be more due to the fact that a wrecked brain description does not really do anything.

Such a destructive algorithm would not be of much help in this argument. If we used such an approach then the same algorithm - one which simply wipes the brain description or otherwise makes it cease to describe a brain that does anything - could be used to correct any problematic behaviour in any brain. The idea is to use the length of the algorithm needed to correct the description as a measure of responsibility and if the same algorithm could be used in every situation then there would be no distinction, with regard to responsibility, between different brains: all brains that committed an act would have the same, minimal level of responsibility for it.

To make it work we need to have greater restriction on the algorithm that "corrects" the description. Instead of allowing it simply to ruin the brain description in the simplest way possible it has to be required to produce what would be regarded as a "corrected" description. The length of algorithm required would then vary enormously for different brain descriptions. I do not propose to solve this problem in this article, though a minimum requirement would probably be that we would expect the "corrected" brain to have certain faculties if it had them in the first place - such as ability to converse in human language. More debate would be useful in this area.

Selection of Language Used for the Description

If the argument is to depend on algorithms that operate on descriptions of brains then someone may well ask what language should be used to formally describe these brains. It would be possible to "cheat" by using a very high level language designed to include some of the information about how to effect a "correction" in the language definition itself.

Rather than get involved with this problem here I merely suggest that a low level language is most appropriate and that I dealt with the issue of what a low level language is in my previous article What is a low level language? [7].

Rationale

I propose such a methodology for assigning responsibility because it seems to fit well with our intuitive expectations.

If someone had behaviour that was easily correctable by medication we would be more likely to consider that person "ill". The medication that we currently have is hardly able to reach into a human brain and individually rewire every neuron to make the brain "acceptable". Any pills that we can give are only able to make a coarse adjustment to a brain - they are not complicated enough to do otherwise - and that goes even for the most sophisticated drugs that humans possess at the time of writing (2006).

If we can correct behaviour with a pill then the intervention that is made by a pill should be something that we can algorithmically simulate. Let us suppose that we take our original brain description and correct it using an algorithm that simulates the administering of medication and produces a description of a "corrected" brain after the pill has done its work.

The algorithm simulating medication in this way would be small - in comparison with many other algorithms that we could imagine. It has to be, because our medication simply does not do things in that much detail. This means that our intuitive expectations seem to fit with the result: simple interventions in the real world are ones which could be simulated by small algorithms applied to descriptions and brains in these sorts of states would tend to be viewed as less responsible.

I am not saying that a condition must be curable by medication. The same argument would assign less responsibility in situations where we had a surgical fix - simply because human surgery is relatively crude. An argument like this could be used to assign low responsibility in situations that are not curable by current medication. In such situations we may simply not have invented a drug to do it yet, but the complexity of intervention required may be the same as what is needed in another easily treatable case.

I am not saying, either, that there has even to be any possibility of a treatment equivalent in complexity to current medication for us to assign low responsibility. Humans may possess nanotechnology in the future. Suppose this was used to correct a situation against which current drugs would have no chance? The fact that a treatment is physically possible does not automatically mean that responsibility is low; however nor does the fact that a treatment requires a more complicated intervention than anything that we can do now automatically mean that there is a huge amount of responsibility. When it comes to modelling such a nanotechnological therapy - in the sense of constructing an algorithm which has the same results on a description of a brain - we may need much more information than we would need to describe the much more simple medication that we have now, but this does not mean there is automatically a lot of responsibility. Even that information may be much less than the information needed to assign a significant level of responsibility.

What I am saying here is that I do not want people to think that the argument here is as simple as "If we can easily treat it with medication then there is low responsibility". The argument is not about specific treatments at all.

When people's behaviour is compromised and we sympathise with them we often view the behaviour as taking over a person and the real person struggling to get out. It is often said that someone "has some good in them". We may think that a person who performs certain acts is desperate not to do them, yet feels unable to stop him/herself. The idea of using the complexity required of a hypothetical therapy as an indicator actually fits in well with all this.

We could debate about what it means to say that someone is having the subjective experience of wanting to do one thing while doing another, but if some part of the person's brain contained, in some sense, the motivation to behave correctly - the information about how to be "good" - then an algorithm performing the correction could read this information from the description and use it to perform its work instead of having to contain it itself - allowing it be shorter and meaning that the person may be viewed as less responsible for actions which he/she, at some level, did not want to perform. I accept that that is a simplistic way of putting it: this article is only providing a basic consideration of the issue.

Scope and Timing of the Description

One issue needing consideration is scope of the hypothetical correction. We should ask how much this imaginary correction is expected to achieve. Is it supposed to prevent one act for which we are trying to assess responsibility, or is it expected to remove a behavioural tendency? My view is that it is probably easier to remain consistent if we are more interested in behavioural tendencies. In such a view we would not really be holding someone responsible for an act as for having a certain type of brain which is liable to perform that act.

Timing is another issue. If we are to imagine a correction acting on a description of the brain, another issue is that of when the description is made. This is closely related to the issue of scope. If our scope were the specific act then we would probably be interested in the brain description at the time of the act, or at the time that it was planned. If our scope were a behavioural tendency then we would probably be prepared to consider the description of the brain at whatever time we happened to be considering the matter - that is to say, now.

We can get an idea of these issues with an example. Let us suppose that a man gets drunk and murders someone. We have him in front of us at some later trial. Some ways in which we could approach the matter are as follows:

1. We could consider the scope of the correction as being the making of the decision to kill someone and consider the brain as it was at the time of the action. If alcohol were almost wholly responsible for what happened and the person would have been considered "normal" without it then it would be likely that little responsibility would be assigned. This is because alcohol is simple - in comparison with many other influences that could affect a brain - and the algorithm to "correct" a brain description of drunkenness would probably be shorter than many we could imagine - maybe doing little more than removing any reference to molecules of the alcohol from the brain description. On the other hand, most people do not commit murder even when drunk. If the alcohol just helped and serious correction is needed of other defects in the brain then greater responsibility may be assigned. Most people would probably not be satisfied if the alcohol consumption allowed much reduction in responsibility.
2. We could consider the scope of the correction required as being the tendency to commit murder and consider the brain description as it is now. This would seem problematic to many people. Now that the alcohol has worn off there may not be much of a tendency to commit murder anyway - meaning that no correction is needed at all. This may assign the person practically no responsibility, but it is more likely that he does have some tendencies to be violent. The alcohol probably had a large influence though, and it is likely that the alcohol effects wearing off would amount to some correction being performed before we need to consider it, so the responsibility would be greatly reduced. Most people would not find a result like this satisfactory. They would say that the drinking caused the murder and he did the drinking.
3. Most people would probably find the previous approach unsatisfactory because the person did make the decision to drink. Whatever caused the drinking in the first place has not been corrected by the alcohol wearing off. Some people would argue that we should look beyond the act itself and consider the tendency that the brain had to make the reckless choice of getting drunk. This demonstrates that the brain we have now - a sober one - has the capability to be reckless in this way. This, combined with any violent tendencies that are likely to manifest themselves while drunk could mean that the current brain description is in need of considerable need of correction. If the algorithm needed to do this is long then considerable responsibility could be assigned.

Of these three, I think that the last one is the best, although it may not be perfect. We have to consider all of the behaviour that the brain is capable of and what it can cause. We cannot just consider the brain's capacity to directly perform violent acts, for example: we need to consider its capacity to do things to itself which increase its capacity to perform violent acts. If we ignored this we could have an absurd situation in which a person could deliberately drug him/herself into a state of aggression which would cause him/her to commit a murder, get the murder performed, wait for the drug to wear off and then claim that his/her brain is in no need of any correction, so he/she cannot be held responsible. Clearly, this would be a fallacy: the tendency to plot the death of other people by taking drugs to reduce responsibility would very much need correction and the algorithm to do this may be very long. This is more consistent. We should not ask if someone will kill "directly" but whether or not they will start a chain of causal events leading to death. Any act of murder imvolves some causal chain, however short. When asking if someone has a tendency to kill we are not really asking if he/she will kill "directly": we are asking if she/she has a tendency to do things to alter reality in such a way as to cause the death of other people. An alteration could be changing the status of a gun from unfired to fired, or the position of a person from near the edge off a cliff to off the edge of a cliff. If the alteration of reality is made on the person's brain itself- for example changing the status of the brain from not drunk to drunk - then we are making a special case of it if we regard that as different from any other change to reality that causes death - and the decision to get drunk is always made while not drunk.

This would make sense in many situations. For example, someone who was involuntarily and temporarily changed by someone else to cause him/her to commit a murder that it was against his/her usual nature to commit would have no responsibility after the change was undone. Most of us would agree with this and assign responsibility to whoever made the malicious change.

It does, however, cause some problems. What if a person's brain is deliberately altered to compromise him/her in this way and the alteration cannot be undone for years afterwards? If the change is by means of something simple and the description could be corrected by a short algorithm then this situation may be viewed as one of low responsibility anyway, but what if the change were one requiring a long algorithm to undo it in the description? Some people may regard the person as a victim, even while he/she continues to have murderous tendencies that require significant algorithmic correction of the description.

A further problem could arise when someone commits an offence and is not caught until much later. In the intervening time he/she may have changed a lot so that he/she regrets the offence and has no tendency to do anything like it again. We could even imagine a person planning to offend and then use some hypothetical technology to alter his/her own brain and remove any tendency to reoffend from it. If we use an algorithmic test on the person's brain after he/she has changed then it may seem that no correction is needed. This could suggest that no responsibility should be assigned - a result with which most people would probably disagree. Some readers may comment that it looks like I am trying to "cure" people rather than assign responsibility, but I am not really trying to do that - we are purely interested in assigning responsibility here and the idea of what would be needed to "cure" someone, in terms of lengths of algorithms needed to operate on a brain description is nothing to do with any desire to correct the situation, but is a standard for assessing responsibility. We are also considering assigning responsibility based on corrections in the past that are clearly beyond us, short of having time travel, so this is not really about therapy. I would agree, however, that this case causes problems.

What this standard provides is a means of assigning responsibility to someone based on behavioural tendencies and what his/her brain is like. In most cases it would be a standard that agrees well with intuition. The two problematic examples here cause problems because we are not just restricted to assigning responsibility at a single time.

In principle, we could get a brain description for every instant of a person's life, determine any undesirable behavioural tendencies that a person has at each instant, and assign responsibility at each instant based on the length of corrective algorithm needed. This means that at every instant of a person's life we could obtain a value indicating how responsible a person currently is for what he/she is. This raises a number of issues:

1. If responsibility were higher in the past, but has decreased now, do we still hold the person responsible? That is to say, do we adjust present responsibility sometimes when we decide that some responsibility from the past should not go away as the brain description changes?
2. Are there situations where a person could enter a state of high responsibility where we may think the responsibility needs some adjustment to take account of some mitigating factor outside the realm of the basic idea of algorithm length?
3. Do we hold people responsible for what they are or for the consequences of what they are? Two people could have the same tendencies to drink and drive for example, yet one of them may be particularly unlucky when someone walks out in front of him/her.

I am raising these issues to show that further discussion is needed; however I think that the standard proposed here has substantial merits. It does at least provide a basic idea of what we may mean by responsibility which would tend to work well in conventional situations. Even if we have to worry about when to measure responsibility, we are in a much better situation than having to rely on totally ad hoc principles.

Idealism versus Necessity

This article has taken an idealistic view about assignment of responsibility, but it may be that more pragmatism is needed in real situations.

Those dealing with people who perform undesirable acts may not need to think about more than just responsibility. They may need to consider such things as deterrence of other people from performing similar acts and such requirements may conflict with idealistic notions of responsibility. It would carry little weight to show that people in some group are not responsible for what they do if it could also be shown that a clear threat would deter some of them from doing it.

For this reason the idea proposed here is better considered as one that could influence decisions about, for example, how a state should deal with offenders, along with other ideas and requirements, rather than as an attempt to propose an entire approach to criminal justice.

Is responsibility a viable concept?

Some people think that if free will is an incoherent concept, then the idea of responsibility should be abandoned. In this article I am not taking a position on that. I am merely proposing that if we have a concept of responsibility, and while we continue to use such a concept, we should seek consistency in how we assign it to people and this means that we should be able to describe what it is.

Conclusion

This article has proposed that we could attribute responsibility to human actions that are considered unacceptable by measuring the length of the shortest algorithm that could process a formalized description of the brain in its current state and produce a description of the brain which is considered to be "corrected" of the behavioural issues.

This would allow a degree of responsibility to be assigned independently of the availability, or otherwise, of any current technology to actually change the way that the brain works.

An alternative method would be to consider the number of algorithms that exist that can process a description of the brain and produce a description of a "corrected" brain.

The issue of what constitutes a "corrected" brain is not dealt with in this article, but even with that as an outstanding issue, having some idea of how to deal with the issue of responsibility, when all human motivation is due to physical causes, is better than nothing.

Actually, accurately performing a test like this is beyond our current abilities. We would not be able to actually get hold of detailed brain descriptions on which algorithms like this could act and we lack any method for producing such algorithms. The test proposed here is really in the realms of thought experiment. This does not mean that it is without purpose. It can still help us to get an idea of what we mean when we say someone is "responsible".

We could also apply a test like this informally. For a given situation we may not be able exactly to perform the test as it is described here, but we could perform some reasoning which would give us a rough idea of what the results would be if we did perform it. As a simple example, I have no concerns at all about saying that a test like this would attribute very low responsibility indeed to the person in the example whose actions are overridden by an implanted computer chip or to a person with some simple chemical imbalance in his/her brain that could easily be remedied with medication. I would find it more likely that it would attribute higher responsibility to someone who murders due to racist views which he/she has as a result of continuous indoctrination by the deliberate provision of hundreds of complex formative experiences in his/her childhood: the tangled mess of causation on which such a person's personality is based is likely to be much more complicated to reverse - with a long algorithm being required to describe such a reversal. To anyone attempting such reasoning for specific cases, however, I would suggest caution.

Some cases can present problems. This is mainly caused by the issue of timing - the standard assigns responsibility for what a person's brain is like at any particular time but there is still the issue of when such responsibility may need to be adjusted to deal with things that have happened in the past.

This does not mean that the standard suggested here is not useful. It is a refinement of the simplest idea of responsibility that we could have, which is that a person is totally responsible for everything he/she does. Such a simplistic view of responsibility needs adjustment in some situations to deal with the idea of diminished responsibility and various "messy" rules associated with it to make such adjustment. The standard proposed here explicitly makes allowance for diminished responsibility and should need less adjustment by special rules.

This article is not intended to be the last word in this matter. It is a very basic consideration of the issue to promote further discussion.

References

[1] Baggini, J. (2004). What's It All About? Philosophy and the Meaning of Life. London: Granta Books. Chapter 7, p118.

[2] Ibid. Chapter 7, p119.

[3] Herring, J. (4^th edition, 2005). Criminal Law. London: Palgrave Macmillan. Chapter 10, pp222-226.

[4] Hofstadter, D. R. (1980). Godel, Escher, Bach: an Eternal Golden Braid. New York: Vintage. Chapter 15.

[5] Web Reference: Almond, P. (2003). Game Theory with Yourself. Retrieved 30 December 2003 from http://www.paul-almond.com/GameTheoryWithYourself.htm.

[6] Zindell, D. (1989). Neverness. London: Harper Collins.
(Originally published:1988. New York: Donald L. Fine Books)

[7] Web Reference: Almond, P. (2005). What is a low level language? Retrieved 17 July 2005 from http://www.paul-almond.com/WhatIsALowLevelLanguage.htm.