Reductionism, Occam’s Razor, and Real Life

Reductionism, Occam’s Razor, and Real Life

I curate a blog (much more nascent than this one!) which aims to reflect how the study of psychology makes our lives a little bit better on a day-to-day basis. I was recently contacted by a reader asking whether the study of psychology was reductionist. It got me thinking – is (and should?) the discipline look for ever more simple explanations for human behaviour? Or is the quest for simple and parsimonious answers doomed to failure?

The problem is that reductionism can lead you to fail to understand the actual nature of the phenomena completely.

Broadly speaking, reductionism is the tendency to understand a complicated phenomena by breaking it down into smaller components and attempting to reduce complex explanations for phenomena down to the simplest possible. This can be useful in that it makes those aspects manageable in terms of studying them. For instance, asking “how a person works psychologically” is a bit daunting – asking how a particular memory system works seems (a bit!) more possible. The problem is that reductionism can lead you to fail to understand the actual nature of the phenomena completely. This could be because you mistake the detail for the picture. I can, for example, understand the mechanisms of how an elephant’s trunk works in incredible detail. If, however, I don’t know what an elephant actually is, I will never really understand the trunk!

Psychology and reductionism

Psychology (particularly empirical psychology) can fall into this trap – attempting to focus on tiny processes or specific issues, without integrating these findings into a whole. Why may this be, and what can we do about it?

As an undergraduate student I was taught the value of Occam’s Razor- the idea that the most parsimonious explanation for a phenomenon (in our case, understanding behaviour) is the most scientifically sound. We can all think of politicians, managers, and adverts which endorse this – reducing complex situations to sound bites. This is reductionism in its purest (and perhaps its most impactful and insidious) form. Some (most, one could even say) psychological research is by its nature reductionist when taken in isolation. Work which focuses on one tiny process in a system (for instance, a small part of the neural system, or a single aspect of intergroup relations) could well be seen as reductionist. However, this work must be reviewed in the context of the field more generally – other people will be looking at other parts of the neural system, others how that neural system maps onto cognition, how cognition maps to motivation and its influence by social systems, among others. Continuing with the animal theme, a nice metaphor here is that of the skin of a fish. If you think of each scale of the fish by itself, it is pretty useless. In contrast, when you take thousands of scales (and understand them as a whole) you get something with full form and utility.

Beyond reductionism

How is psychology doing on this front? We are a relatively young science, but I think we are amassing a pretty good collections of scales already! However, we still need more eclectic models to move beyond reductionism. There is, for me, one key thing the field needs to get better at –  we need to start generating multi-level explanations which understand things from the micro (i.e. neurological) to the macro (social systems) level. Initially at least, a single model cannot account for all of these, but a number of parallel models with clear links could almost as well, and would be a welcome start. To achieve this, we need to be more aware of the fact that our own take on a subject represents a single perspective, and be more active at trying to integrate our own level of explanation with others work.

I hope my own work is beginning to reflect this multi-level approach. In my research programme (around the role of social identity on a variety of behaviours and outcomes, but focusing on recovery from addiction), I have become increasingly suspicious of simple solutions or explanations. Sure, a given variable (say, levels of identity as a recovering addict) may predict a lot of variance in another (say, relapse rates). I could conclude that this is the simplest answer, and run with it. But I also know that a whole bunch of other stuff is going on – in my example, some people are new to quitting whatever their problem behaviour was while some are on their nth attempt. Some have supportive families, others are surrounded by others who engage in and encourage the problem behaviour. As a result, my empirical data (and the theory which underpins it) gradually gets more and more complex; I add caveats. The interactions between different factors and the moderations and mediations between them become as interesting as the original relationships. Increasingly, I see things acting as gestalt systems with effects due to complex, emergent and, often, messy causes. As my work progresses I find my explanations, contrary to reductionism, getting more and more complex.

I’m also trying to use more investigational tools to look at things from different levels.  I am interested in the implicit (automatic, unconscious) nature of identity, and explore such effects by using reaction time tasks and cardiovascular responses. These pin down effects to specific mechanisms – a nice, simple, Occam’s-Razor-Compliant set of processes. But I’ve also started dabbling in qualitative methods – interpreting the conversations recovering addicts have with interviewers. These throw up all sorts of themes, systems and randomness. The experimental psychologist in me sees this as unhelpful noise in my data. In contrast, the newly emerging multi-disciplinarian in me sees a whole new set of questions and ideas to explore!

To conclude, is the discipline reductionist? I think you could see it that way. However, there is sufficient diversity in the field (in areas of interests both overlapping and disparate) that we should generate the insights we need – we “just” need to  integrate them . We also have such a variety of methodological tools at our disposal (ranging from qualitative interviews to MRI scanners) that I suspect (hope?) it won’t be long before multi-level explanations become the norm, and “meta-theories” encompassing them become more available to work with. In the meantime, I’ll try and see the elephant on the trunk, and the fish not the scales!

Daniel Frings is an Associate Professor at London South Bank University. He has published research extensively in various fields including decision making, performance, health and social influence. His particular interest is in social identities and motivational processes. Alongside his academic work, Daniel also curates Psychology It Bettera blog which aims to show how psychology can improve our everyday lives in ways both large and small.



Share This Post

One Comment - Write a Comment

  1. How synchronistic you mention fMRI where Bayesian statistics is employed to develop the actual picture from the raw data vectors. And I was just talking this on a blog here that was trying to go beyond the way we reduce to simple true false statements instead of multiple probabilities, most to least plausible. and how mind might use such methods. It all ties back to Gestalt, of course.

    “Basically, Gestalt therapy is concerned with the interaction between the organism and its environment (Baumgardner, 1975). In the healthy organism, many needs are present at any one time. These organise themselves into a hierarchy of importance, as a natural process. The most dominant need forms, or becomes figure. In order to satisfy this need, the organism searches its environment for the desired object (sensory activity); when the object is found, the organism acts to assimilate it (motor activity). The concept of the connection between the sensoric and motor activities has been extensively discussed by the Gestalt psychologists for many years (for example, Koffka, 1935; Hartrnann, 1935). When the needed object has been assimilated, the Gestalt is closed, and a state of equilibrium is reached. The formerly dominant need recedes from awareness (becomes ground), and the energy thus freed is directed towards the next most dominant need. Organisms are thus self-regulatory or homeostatic. (Smuts, 1926; Lewin, 1935; Goldstein, 1939; Latner, 1973; Perls, 1973). In this way, the organism is regarded as being “born with the capacity to cope with life” (Simkin, 1976, p.17) Retrieved from

    Signal? Noise? we are learning how to pluck the signal and as to my contemplatively digested conclusions to date…we are finally on the right track, with the right tools to see some true progress. And the experimentalists will be there with you, reformed of their sins. labs are necessary, but not sufficient.


Leave a Reply