[extropy-chat] Neural Engineering

[Anders Sandberg]

fredagen den 19 mars 2004 06.10 wrote Robert J. Bradbury:
> Now what I am interested in from the perspective of the
> "Neural Engineering" perspective is the possibility of
> extracting (non-destructively and non-painfully) information
> that leads directly to the people who are responsible for
> the direct (and most probably painful) termination of
> innocent people.

It seems that the question here is how strong our right to our own mindstates 
is compared to other rights. This is actually a good question.

As I tend to think of it, we own the contents of our minds just as we own our 
bodies. We have the sole right to determine what to do with our internal 
information, whether to reveal it or remain silent. This right appears to be 
stronger than just the ordinary ownership rights just as the right to one's 
body is stronger than ordinary ownership of external objects. 

The freedom to choose is part of this; the ability to make one's own decisions 
is not just tightly linked to the mindstate but also to being a moral subject 
at all. If that is infringed it is even more serious than just 'stealing' 
someone's mental information. But the current scenario is not about limiting 
this freedom, just checking that the mind does not contain any plans for 
killing someone, or the memories of having done so.

My basic heuristics of what rights may be infringed on in the pursuit of 
justice is that only weaker rights may be violated when trying to deal with a 
rights violation (e.g. stealing (right to property violation) should not be 
punished by death (right to life violation)) and that coercion may not be 
initiated. Being a minarchist, I acknowledge that we can delegate a monopoly 
on coercion to the (local) state if we are careful. 

>From this standpoint it is entirely OK to make a social contract where the 
police is allowed to check the minds of people in the pursuit of a crime. It 
is not truly different from all the other investigatory powers we allow the 
police, powers which we accept since they produce a safer society (assuming 
the police to be competent, not corrupt etc). The integrity violation is 
lesser than the life violation done by the murderer. Such scanning is only 
moral if the crime is at least a crime against the right to one's body or 
mind - no scanning for thieves. But these investigations are after the fact: 
scanning brains for murderers is OK, but it is not clear it is moral to look 
for would-be murderers. 

In current legal practice it is often problematic even to look too broadly for 
suspects, there has to be a reasonable amount of evidence to allow invasions 
of privacy. One could of course assume a kind of transparent society 
contract, where random scanning for serious misdeeds (murder, rape, 
brainhacking) is accepted. But we are a long way from that, and building 
institutions that could handle that power without leaking or corruption is 
hard. 

It is the would-be murderers that are problematic. Innocent until proven 
guilty is an important heuristic for open societies - it embodies the trust/
niceness part of the reciprocal altruism we need to keep the society 
functioning, and it shows that we value freedom higher than punishment. This 
means that coercive scanning for potential crimes is not really acceptable, 
it violates both the non-initiation of force and produces violations of 
rights. Doing a voluntary scanning showing that one is "nice" is of course 
OK, and can be included into making voluntary contracts. It is the forcing of 
scanning that makes it problematic, and this is compounded by the idea of 
pre-emptive justice. Pre-emption places the burden of proof on the accused 
("Prove that you *won't do it!") and tends to lead to unstable situations 
where it is better to act before anybody else has a chance to act, producing 
rash decisions. 

Would the benefits of finding these dangerous persons outweigh the risks? I 
doubt it. Assume that one would-be murderer in 10 actually does murder 
someone (the exact numbers are of course uncertain, but 1:10 doesn't sound 
that strange - it could be 1:100) and one in a hundred is a would-be murderer 
(probably a bit too high :-), and that we use scanning on the entire 
population. That means that 99% of people get their mental privacy violated. 
Then we have the 0.9% potential murderers we now have to deal with. Most are 
simply in need of some help (psychological, economical, social, whatever). If 
they all can get it it isn't that bad, but this is very unlikely to happen in 
the near term. More likely they are treated in ways that limit the risk to 
the others, which in general is a limitation for them. So we end up with 0.1% 
stopped would-be murderers, and 0.9% people who actually would never have 
murdered anybody but now have their freedom circumscribed anyway. Plus the 
basic 99% of coerced citizens. Is the price of preventing 1 murder worth 
putting 9 innocents (possibly nasty people, but innocent) in custody or 
permanent monitoring? (plus 990 people who got privacy invasions). I think 
most people disagree. 

Maybe if that single murderer really was so dangerous that his damage was 
"worth" 9 innocents in jail, but that leaves only some of the nastier 
terrorists. While it is hard to compare rights (they are qualitative things), 
one could perhaps view them as orders of magnitude of utility. If a wrongful 
incraceration is worth 1/10 of wrongful death and privacy invasion 1/100, the 
total "cost" of the above scheme would be about 10 lives - and that is *real* 
lives, if we are looking for risks it must mean that the probability of the 
crime times the number of lives lost > 10 (assuming the above 1000 people 
scenario). Which means that we need to be pretty certain (more than 1 chance 
in 100 of the crime actually happening) even when dealing with a potential 
~1000 victim attack. 
[I think the reasoning in this paragraph is seriously flawed, but maybe one 
could make something better out of this mess. The assumptions of the "sizes" 
of the rights is downright random. ]

To sum up, I think looking for would-be murderers using neuroengineering is a 
dangerous step both morally and socially. Using it to find murderers is far 
more OK, we only need to set up proper safeguards about the information and 
instutions handling it. 

> There is a converse side of this perhaps -- i.e. those
> individuals/companies who offer full disclosure ("go ahead
> read my mind") so that it is completely obvious that they
> are dealing from an up-front perspective.  

I ran a rpg scenario where one culture (of course, it was a 
libertarian-transhumanist planet) had something like this. Everybody was 
wearing wearables that contained micro-fmri that allowed them to display 
their mental state in augmented reality. Not exactly truth machines, but 
enough to give a sense of the mind behind the poker face. Not showing a 
mindstate was a sign of untrustworthiness, and faking it was a serious social 
gaffe (making sudden, unexpected remarks and checking that the response was 
plausible had become a part of social interaction; offworlders found the 
atlanteans annoyingly rude). 

It is still mostly a symbolic sign, just like shaking hands to show there is 
no weapon there. But it helps a bit.

-- 
Anders Sandberg
http://www.nada.kth.se/~asa
http://www.aleph.se/andart/

The sum of human knowledge sounds nice. But I want more.