A Theory Problem
When we’re thinking about why the replication crisis occurred, it’s easy to focus on the pervasive methodological and statistical issues in psychology. From the overabundance of college student samples to the questionable research practices used in analyzing datasets, there are plenty of reasons to blame the methods of psychology for the crisis. But, as a series of other researchers have noted– there’s a deeper problem lurking here, one that runs back to the beginnings of psychology. Psychology has no theory.
You might think that’s a patently false claim. After all, just within social psychology you can find: Tend and Befriend Theory, Accessibility Theory, A Theory of Impulse and Reflection, Construal Level Theory, An Attribution Theory of Motivation, A Theory of Social Information Processing, Balance-Logic Theory, Lay Epistemic Theory, A Theory of Heuristic and Systematic Information Processing, Feelings-as-Information Theory, Action Identification Theory, Social Cognitive Theory, Cognitive Dissonance Theory, Terror Management Theory, Self-Determination Theory, The Theory of Planned Behavior, Social Comparison Theory, Regulatory Focus Theory, and Mindset Theory of Action Phases. And that’s justpart of Volume 1 of the Handbook of Theories of Social Psychology. If there’s anything psychology doesn’t need, it’s more theories.
But, say you’re a religious studies scholar and you’re interested in the way psychology shapes the relationships that grow in religious communities. Maybe you’ve run into lots of interesting evidence that this social dimension of religion has consequences for health and life-satisfaction, and want to know why that’s happening. Psychology must have something to offer.
When you go to PsychInfo, however, you’re confronted with a dizzy array of theories about the psychology of relationships. Should you use Attachment Theory or Sociometer Theory or Self-Determination Theory or Social Identity Theory?
Maybe I’ve just cherry-picked one topic that’s particularly complicated. But, look through that initial list again. If you’re interested in information processing, there are a half-dozen relevant theories there to choose from. If you’re interested in self-esteem, you’ve got another half-dozen. Personality? A full baker’s dozen (just not the Myers-Briggs or Enneagram please).
For any psychological process or phenomenon, you’ve got a zoo of different theories, each attesting to tell you the truth about that thing. How are you supposed to decide between them?
There’s no overarching theory to help you sort out which is most trustworthy and which is most relevant for your question. The sub-theory you end up using is probably just a matter of which you encounter first or which you’ve heard about.
Guiding new comers to the relevant information isn’t even the most important role of an overarching theory. As psychologists we need some agreed upon framework to sort through disagreements between these different sub-theories. This isn’t to say that all those theories about relationships are in conflict with each other. Most likely the differences between them simply come from their different perspectives. There’s lots of space for integration, but there’s no organizing framework to guide such an integration.
An integrative theory for psychology would also help efficiently generate predictions and sort out those predictions that are nonsense. And perhaps most importantly, if you run an experiment and get a wonky result (like the study that set off the replication crisis in which Bem thought he’d shown people anticipating future events they couldn’t have known about), then a foundational theory would let you know that the result was a mistake.
How do we solve the theory problem?
Fortunately, I’m not the only one talking about this need for theory in psychology. Muthukrishna and Henrich wrote about this last year. Smaldino made the same point three years before that. And Borsboom three years before that (and there are plenty of others making similar arguments).
Despite the consensus around the need for theory, there are open questions about what an integrative theory in psychology should look like. Some of these other researchers are arguing that what we really need are better models. If we had better models then the overarching theory problem would sort itself out.
“Models” can be a confusing term, but it’s just a fancy way to name the formal descriptions of whatever it is researchers are studying. F = ma, that’s a model of physical force. rb – c > 0, that’s a model of inclusive fitness. If math isn’t your thing, think about scientific models like blueprints for a building. The only difference is that the house is already built and we’re trying to draft the blueprint in reverse.
But really, even if math isn’t your thing, in any scientific endeavor mathematical models are the good stuff. This is because they don’t leave any room for ambiguity. As a researcher, if you’ve got a mathematical model for whatever it is you’re studying, then you’ve got a precise layout of the relevant variables and how they relate to each other. This lets you make very specific predictions and makes interpreting your results very clear.
No doubt, psychology would benefit immensely from such models. Hell, it’d be transformative if researchers just attempted to spell out their verbal models in some rough mathematical form. But, in order for us to get accurate and useful models of psychological processes, we also need a higher-level theoretical framework that’s going to involve some concepts that aren’t easily modeled.
The value of vagueness
So much of psychology depends on vague concepts. By vague, I mean the sort of categories that can be specified by contradictory instances. For example, the concept of “religion” is vague: religion can be modeled as a behavioral system of social signals of solidarity or as protests against the prevailing social conditions. Such ambiguity may seem anathema to inquiry, which is supposed to clarify concepts. But the diversity that exists under the vague umbrella concept of religion isn’t just job security for religious studies scholars, it’s also really fruitful for inquiry because it allows for meaningful comparisons.
Returning to psychology, consider the concept of self-regulation, which describes the way that we pursue different goals. This concept is vague because sometimes acting in accord with a goal is a conscious, effortful, and very deliberate process. Other times we self-regulate without any difficulty at all– we’re not even aware that we’re doing it. We can model these behaviors, indeed self-regulation is probably one of the better modeled phenomena in psychology, but as that model becomes more precise should it eliminate this vagueness? Should it become two separate models?
Something crucial is lost if we rush to create precise models at the expense of these vague concepts. The ambiguity within the vagueness that surrounds “self-regulation” is a well-spring of generative debates about the nature of goals, the type of intentionality that we use in different circumstances, the limits of our will, and the nature of unconscious processes.
The same point holds for other vague concepts in psychology like perceptions, cognitions, emotions, defenses, motivations, etc. We study these things, we debate about them, and we try to put them under the lens of empirical scrutiny. The vagueness of these concepts is incredibly fruitful. It’s also honest, because when it comes down to it, we still don’t know what type of stuff these concepts are.
I realize this is a fine line to walk. I started out by saying that we need an overarching theory in psychology, but now I’m saying that we should restrain ourselves from preemptively developing precise models of psychological processes. Which is it?
The point I’m trying to make is that we need a theoretical framework that extends beyond our models. We need such a framework in order to help us sort through which psychological processes are able to be mathematically modeled. We need it to help determine what sort of mathematics is appropriate to represent these processes. And we need it to point out the limits of our representations. Such a theoretical framework should be able to accommodate both vague and precise concepts, and it has to help us navigate the transition between these levels of analysis.
In the absence of such a framework, our rush to formally model the mind risks prioritizing those processes that are amenable to mathematical modeling and obscuring those that currently aren’t. If a collection of such models becomes the overarching theory of psychology, then what wild and novel parts of ourselves are we excluding from analysis?
I don’t want to deny the importance of models on the theory front. Given the near absence of such models in psychology, any should be welcomed. I also don’t want to set limits on our ability to model psychological phenomena– I’d love to see the day when we understood our psychology with the same rigor and depth that we understand relativity. But, in order to get to such a place, we have to be honest about where we currently are and we have to hold our vague concepts with an open hand, accepting their ambiguities and letting those contradictions be guides for inquiry. If we can walk this middle ground between upholding the ambiguities of vague concepts, while also working tireless to specify them, then we can ensure that whatever foundational theory we generate will be integrative enough to trust as a way to understand ourselves.
Written by Jonathan Morgan, Postdoctoral Fellow. September 3, 2020.