Agents, Mechanisms, and Self-Deception

Hi everyone--here is a last post for January before I pass the baton to Derk. Thanks again to Thomas and to all of you for such thoughtful and interesting contributions--the whole experience has been terrific, and you’ve given me much to continue to think about!

One very influential approach to responsibility seeks to understand it in terms of reasons-responsiveness.  But there are a variety of ways of understanding reasons-responsiveness, and these include mechanism-based views and agent-based views.  So, for example, on Fischer and Ravizza’s mechanism-based approach, one is responsible when one acts on one’s own moderately reasons-responsive mechanism.  Without getting into the subtle details of their account, we can understand a mechanism to be reasons-responsive in general as follows: if the mechanism on which one acts were to operate, then in some set of worlds in which there were sufficient reason to do otherwise, the agent recognizes the reason and does otherwise.  Here it is the mechanism on which one acts that is reasons-responsive in the first instance.  In contrast, on an agent-based account, an agent is responsible for an action if the agent herself is reasons responsive.  (Reasons responsiveness on the part of an agent can be cashed out in different ways, including as having the ability to respond to reasons that there are in the situation.)  Much has been said about the contrast between these two different approaches, but having recently been thinking about self-deception, I’d like to come at it here from perhaps a slightly different angle. 

Consider a case that allows for an agent to be reasons-responsive even when the agent does not act on a reasons-responsive mechanism (understood as above).  Suppose that a woman deceives herself, coming to believe that her son is not abusing his own children, for example.  Suppose the mechanism on which she forms the belief--or collects her evidence--is a motivated biasing mechanism.  Supposing that the biasing mechanism is not reasons-responsive (or does not meet the requisite threshold level of reasons-responsiveness), then the self-deceiver will not be responsible on the mechanism-based view.  In contrast, on the latter view, it is not exonerating that a non-reasons-responsive mechanism is operating.  What matters is whether the agent could have either prevented that mechanism from operating, or instead put another into action.  Since there is good intuitive support for the idea that self-deceivers like this woman can be responsible, does the case of self-deception provide support for the agent-based approach to responsibility?   

Comments

[Neil Levy]

I'm having trouble understanding the supposition. Is a "motivated biasing mechanism" a mechanism that has function of biasing reasoning? Why would we have such a mechanism? How would it evolve? It seems more likely that motivated reasoning is the product of an interplay of mechanisms: independent signals of costs alter the workings of whatever mechanism is involved. Neuropsychologically, almost all reasoning is the product of interacting mechanisms. It is open to Fischer to identify the relevant mechanism with the set of (neuropsychological) mechanisms that are involved in a token piece of reasoning, though I suspect that would entail that no mechanism is actually moderately reasons-responsive (because if we hold fixed that mechanism, we will see some weird reasoning).

[V. Alan White]

One question Dana--what constitutes the differences between the abilities of R-R mechanisms and agents? (Say dispositional analysis versus some categorical one.) Is there some deeper issue of what constitutes ownership here? And thanks for a great month of Q&A on flickers!

[Philip Robichaud]

Dana,

You said that agent reasons-responsiveness view gets this case right because:

“The agent could have either prevented that (non-reasons responsive) mechanism from operating, or instead put another into action.”

Can’t the proponent of a mechanism-based view of reason-responsiveness respond in a similar fashion? Let’s grant that the mechanism on which the self-deceiver formed her belief wasn’t reasons-responsive. Still, the self-deceiver’s responsibility could be established by tracing to a prior mechanism that was moderately reasons-responsive. For example, imagine the woman knows that she will see her son in better light if she looks through a particular photo album that depicts him as a loving father. Just as she starts to suspect that he may be abusive, she chooses to look through the album, and, boom, she maintains her belief that he’s a great dad. We can imagine that the mechanism on which she chose to look at the album was moderately reasons-responsive – there are worlds where she would not have made that choice. Importantly, the effect of her choice is the enlistment of a non-reasons-responsive, (b/c biased) evidence acquisition system that leads her to form the false belief. Thus, her responsibility (for subsequent ignorant action on that false belief) might turn on the existence of a trace to a prior reasons-responsive mechanism that could have prevented the downstream non-reasons-responsive mechanism from operating. Of course this brings with it a commitment to tracing, but that might be something the proponent of mechanism reasons-responsiveness might be willing to take on.

[Bruce Waller]

Hi Dana, Neil, Alan and all,
First, an echo of Alan's thanks, it has been very enjoyable, as well as giving better appreciation of some of Dana's rich and detailed arguments in her book. On Neil's question, would a "motivated biasing mechanism" be functionally defined, perhaps taking several different forms? I'm thinking of your very insightful discussion of racism in the "non-culpable ignorance" section of HARD LUCK, involving self-deception. But that brings up a difficult question -- because as Neil notes, agents do not control self-deceptive processes; so how could an agent be responsible for changing that process, or for switching to a better process, or recognizing the need to switch to a better process? Does that raise doubts about the adequacy of an agent-based solution to this problem? Again, thanks for the excellent discussions.

[Dana]

Lots of great issues raised here.

Neil raises the question of what the mechanism in question is in a case of self-deception. I wouldn’t want to try to answer the questions of whether or why or how such a mechanism would evolve. But I think it is ok to set aside evolutionary considerations (and in any case similar questions could arise for the interplay of mechanisms as much as for mechanisms, I would have thought, insofar as I understand them). Still, Neil makes available a nice reply on behalf of the defender of the mechanism-based view (before giving his own reasons it won’t ultimately work), and that is that I’ve incorrectly individuated the mechanism on which the grandmother acts. I’d be interested in seeing whether this kind of reply could be developed. In the meantime, it seems to be at least a coherent (if not empirically accurate) story that the grandmother acts on a non-reasons responsive mechanism (whatever it is), and yet could have acted on a different one instead. And perhaps this is all that is needed to provide some support for an agent-based view.

Philip offers a complementary defense of the mechanism-based view here by appealing to tracing. The idea is that the grandmother can still be responsible even though the mechanism is not reasons-responsive if she earlier acted on a reasons-responsive mechanism, in such a way that the activity is connected in the right way with her currently acting on a non-responsive mechanism. That’s also a neat strategy. Here’s one reply: I don’t think all cases of self-deception for which we are intuitively responsible can be accounted for by tracing considerations. (Of course, as Philip notes, it would help to have a worked-out account of tracing, too.) Suppose that the grandmother doesn’t do anything that satisfies plausible tracing condition earlier--there was no point at which she could foresee, or should have foreseen that her earlier choices would cause her to act in this way, etc. But suppose she could make an effort now to face the evidence and does not. Would we now think she is not responsible because she does not act on a reasons-responsive mechanism?

Bruce and Alan raise two deep questions I had so far managed to avoid--Bruce’s about the very nature of self-deception and Alan about the nature of agential abilities. Perhaps my earlier reply to the tracing defense offered by Philip can speak to Bruce’s question. I grant that in at least an important sense we do not (typically, anyway) control the processes of self-deception, and my own view of self-deception is that no intentional act on the part of the self-deceiver is required (in this I believe Neil and I are in agreement). But from this it does not follow that we lack control over other processes that might undermine the continuation of that process. As for Alan’s question, that’s a hard one! So the short answer for now--in the book, I develop a non-dispositional, but nevertheless compatibilist, account of the agent’s abilities. But for the moment I was hoping to be neutral in thinking about the case.