These are my notes of the discussion during the third European workshop on causal reasoning in clinical decision making. The workshop took place on April, 25th and some experts from different European countries were present. My notes are not at all complete, they were rather meant to help me remember the interesting points that were discussed during the whole workshop.
In Belgium a program was designed that uses specific rules to compute reasoning from a questionnaire of Likert scales. That did sound really interesting and I would like to have a go with this program. It could also be very interesting to make an online version and get a lot of people to fill it in, so we have a norm-group.
Differentiate the causal models
During the workshop people used different causal models (classification , explanatory…) in their presentations. It would be prudent to set those causal models apart and look in what ways they are different/the same. This would also make it easier to compare the research presented.
Discuss function analysis with your patients
Actually THEY are the best knowledgeable expert about themselves, so discussing the function analysis can tell you a lot more, because you can see how they react and if what you got seems to be true. Maybe you wondered off a strange path…(still be careful about that – they might also want you to follow a specific path).
“Causal” is a difficult term in clinical therapy.
Actually the therapy can be seen as a way to test the causality-hypothesis that you have come up with. But still then causality on psychotherapy is different form causality in physics. Also worth thinking about: Does your believe of causality change if the therapy fails?
Research in the field
With regard to the research in the field three final points were discussed. First it was mentioned that to the your hypothesis it would be prudent to construct critical cases. You really need to know your theory to do that because you will need to find the critical point in which you expect the reasoning of the two conditions to differ. Secondly you should take care at what you are aiming at: Is it a explanatory model or a treatment model? Methods are quite different… Finally Nadine suggested that the level of abstraction is also an important factor. Looking at cheater detection research for example, people are quite bad doing this in an abstract context, but are very good if social rules are used.
Alright…that’s it. See you all next year in Spain (-: