This is a very interesting question, and one that I have been wondering about recently. From my point of view, it is not the mechanistic methodology that is the problem (although this may be problematic if one assumes that all psychological constructs one measures are necessarily mechanistic in nature), but rather the fixation on reductionism. Reductionism leads to a explanatory 'race to the bottom', in which behaviour is to be explained by cognitive entities (representations), which are to be explained by neural processes (usually regions of activity), which are to be explained by chemical reactions, and so on until psychology becomes the study of molecular behaviour. This reductionist goal of explanation of course influences methodological decisions, in which potentially interesting behaviours or situations are stripped away to supposedly get at the 'true' processes involved. The danger of this is that psychology becomes the science of constrained and artificial reflexes and their supposed lower-level 'causes' (John Dewey picked up on this point in 1896).

For what it's worth, I try in my head to conceive of psychological experiments as about describing the behavior of interacting systems. This would apply as equally to an intentional moving agent interacting with its auditory environment, as to a motor control region of the brain interacting with changing neural signals from a different region, as to one social group interacting with a different social group in a particular contextual constraint. This approach deliberately eschews the need for reductionism as an explanatory goal, and also allows that interacting systems may be described (measured) in different ways without changing the fundamental thing happening. This helps a little I think, although the real challenge of psychology from this approach will be to explain how different systems relate to each other, and to be able to predict the descriptive outcome of a certain systemic interaction, i.e. to offer a constructive prediction of human behaviour.

What do you think?

I think strategy to achieve useful research for society, must ensure transdiciplina and the simultaneous use of abstract and general approaches, with the contrast of more specific studies (micro) .

Matthew - I agree that systems level thinking it terms of design will probably get us farther, especially since many of the questions "worth" answering are complex in nature and can only be appropriately address at a higher (aggregate) level.

Federico - I like the idea of mixed methodology/discipline approaches, the sticking point seems to be in finding the proper venue for the products of these types of collaboration, given the specialized (reductionist???) nature of most journals and professional meetings...it's not just doing the work, it is reaching the audience. Lots if issues here - but thanks for your responses!

Scientific theory is properly a description of the state of nature. Statistical analysis merely tells us how well a putative description accords with observation. The problem is not ‘mechanism’ or ‘reductionism’ per se, but the incorporation of beliefs (of which ‘mechanism’ and ‘reductionism’ are arguably two) with the observations that scientists attempt to describe. To give a psychological example: It took 100 years from Fechner’s time for it to be shown that the nineteenth century idea of a psychophysical threshold was no more than invention. That was a real breakthrough, but, notwithstanding the seminal work of Swets, Tanner and Birdsall (1961, Psychological Review 68, 301-340), the belief persists to this day (Laming, 2013, J. Opt. Soc. Am. A 30, 300-315). Statistical methods ought to effectively exclude beliefs from the reckoning, but, as you say, that is not how research is actually done.

Donald - thanks for your response, this is a point that I try to relate to my students (although probably not as well as you stated it) - the idea that assumptions and beliefs about the way 'science' should be done is both a necessary constraint (gives us a framework to work within) and a hindrance (limits the type of inferences/conclusions that can be made). I think this is especially true of psychology since from the start the relevant phenomena (thoughts, consciousness) seemed to operate outside of the borders of mechanism/reductionism. Over the years we have found a way to make psychology "fit" better with this approach (speaking as a card carrying experimental psychologist), but there is still a disconnect (hence the original question).

Hi Leonard,

Thanks for this question, with me it certainly rings a bell or two.

In HCI, as you must know, design is considered all-important and pervasive, just because it often seems impossible to get the psychology right. Sometimes a kind of applied psychology helps, as exemplified by the work of Jaob Nielsen and his newsletter, that I admire, but for designers that is no guarantee, they can be stuck in the mud for long periods and can become blocked too, seeing no way out.

When you try to study learning, as I have to because I design speaking communication aids intended for people with limited cognitive resources (due to stress and illness), it can also be difficult not to 'coach' people through tasks, thereby spoiling your own method. This as I get it, is your point.

I think there is no 'silver bullet' here that would work in all circumstances and for each and every problem and/or would reduce all complexity without doing unneeded collateral damage. But it is good to discuss these things and sometimes, in my experience, methods such as 'thinking aloud', 'design rationale', 'argue pros and cons', and most of them the teach back protocol, that is dear test subjects, please explain each other, or explain me, what it is that you learned. Using all your senses. Asking, for example, do I get you right that .... and ... tell me why, please, and may i show you a drawing, please help me understand this....

It will not help much to get things published, because often enough referees do not even get the point of why you want to use non-standard methodology in the first place, but it just might help to find the answers that you are looking for.

Hi Joris - Thanks for your response, In a course I co-teach on creativity, innovation, and design (separate from my research design course) - We try to present a number of ways for students to approach design, from the psychology standpoint I lean a lot on Don Norman, Kim Vicente, and Jacob Neilsen - mixed with with my perspective as an Ecological (Gibsonian) psychologist. From the design standpoint we have been using the IDEO approach mixed with more traditional research methodology to try to provide students with a "toolbox" rather than a "way" - I think this has been successful because it keeps students (and us) from falling into a particular process which can make finding novel solutions difficult.

Hi Leonard,

It seems to me that much of that is not 'real' psychology research?

I would argue that just about anything that affects how we behave could be considered psychological - but I understood your point. I think what my work in more applied settings has lead to is a more systems level approach to research (as was mentioned in a few of the earlier responses) - both in terms of my more technology oriented research and my basic interests on the regulatory influences of perception on behavior (and the opposite). It means that the research designs themselves are a little more complicated and the data are a little more difficult to analyze, but the interpretations and conclusions seem to be a little more relevant. I think all of this raises the question of what are we really trying to learn/discover (particularly in psychology). Food for thought...