Archetypes are a Poor Primitive for a Theory of Mental Representations

Abstract

Few thinkers have entered our collective consciousness as much as Freud and Jung. Most modern psychologists could only dream of such staying power. Becker and Neuberg argue that Jung’s plumbing of the human mind and his description of archetypes may be more than Western woo. They draw on cognitive, evolutionary, and developmental science to make their case. Here we tackle two questions: (1) Were Jung’s archetypes correct in some way? And (2) Are Jungian archetypes a useful theoretical approach? Our answer to the first is possibly, and the second, probably not. To this end, we present an alternative approach to developing a theory of mental representations. Jungian archetypal representations such as the Child and the Caretaker or Leaders and Followers do seem to prima facie map onto fundamental adaptive motivations of mating, childcare, and social status. And some archetypes are more relevant than others during different stages of development, such as the Caretaker during infanthood and the Mate and Rival post-puberty. As Becker and Neuberg argue, such archetypes are useful labels for these cross-cultural, and even cross-species, aspects of life. Various subdisciplines in the psychological, behavioral, and biological sciences have uncovered details about these developmental domains and stages that may have led to reliably developing representations for various roles, relationships, and patterns of behavior that match these particular challenges. So in some sense, Jung was onto something in his suggestion of fundamental archetypes, and these arguably map onto our emerging scientific understanding. What is less clear is whether Jung was somehow more perceptive in his identification of these patterns compared to other philosophers and cultural commentators. Can a Jungian approach provide more insight or are the archetypes themselves tapping into something more fundamental than Seneca (Seneca & Campbell, 1969), Marcus Aurelius (1942) or other Stoics on moral psychology or the psychology of relationships? Or Sun Tzu (Sun, 2017) or Machiavelli’s (Machiavelli, 2018) discussions on cooperation and conflict? Or in modern times, Tobias’ (1993) 20 Master Plots or even the crowd-sourced and eerily accurate TV Tropes (tvtropes.org) that pervade our modern storytelling? One could make a similar case that these examples also provide a window into human life with an ontology that maps onto our scientific understanding. But while Jung and these other thinkers may provide inspiration or even insight, they are an unprincipled and poor primitive for developing a theory of mental representations. This is especially apparent when compared to alternative approaches that build theory from first principles. The idea that genetically evolved biases channel our mental representations toward particular forms that correspond to recurring socio-ecological challenges is a reasonable proposal. Particularly so if the alternative is a ‘blank slate’ argument that insulates learning processes from any content-biases that can be shaped by forces like natural selection. However, there are many theories that explain the same phenomena, including the recent cognitive, evolutionary, and developmental approaches that Becker and Neuberg map onto the Jungian archetypes. It’s not clear what a Jungian perspective predicts on top of these theories, beyond identifying some ambiguously specified psychological domains with little precision, or merely suggesting the existence of reliably emerging motifs. As an example of an alternative, more mature theoretical approach that goes beyond the Jungian thesis, consider Dual Inheritance Theory (Boyd & Richerson, 1985; Cavalli-Sforza & Feldman, 1981; Chudek, Muthukrishna, & Henrich, 2015; Henrich, 2016; Russell & Muthukrishna, 2018). As Muthukrishna and Henrich (2019) argue, the use of formal theory and theoretical frameworks that connect and build on other theories, themselves built from first principles and grounded in the models of evolutionary biology, allows for more precise predictions and additional constraints on both the questions we ask and the way we answer them. They allow us to tackle science as an abductive challenge and move toward more general theories of human behavior. A seminal theory in the Dual Inheritance Theory framework is captured by a model of when natural selection favors social learning over both genetically encoded solutions and trial and error learning (Boyd & Richerson, 1985; expanded and built on by others, including Aoki & Feldman, 2014; Hoppitt & Laland, 2013; Nakahashi, Wakano, & Henrich, 2012). This autocorrelational model explores how the strength of environmental similarity between generations affects the solution space. To summarize the gist of the predictions, when the environment is highly stable, phenotypes encoded in genes provide the most efficient solution to these long standing problems.