| 520 | |
▼a Much of contemporary cognitive science is rooted in the information-processing models of the late 1970's, which envision cognition and action as falling into two categories: goal-driven and automatic. This dual-process perspective replaced the once-dominant view that all cognition and action is preceded by and under the control of a hierarchy of goals. I suggest that the original, "goals-first" framework was right all along: all thoughts and actions, including those that are automatic, are goal-dependent. This idea is elaborated in Chapter 1, and then defended in Chapters 2 through 7.In Chapter 2, I defend the goals-first framework by opposing the dual-process framework that replaced it (Melnikoff & Bargh, 2018). Specifically, I show that the dual-process framework lacks empirical support, contradicts well-established findings, and is incoherent. In Chapters 3 through 6, I provide positive support for the goals-first framework with two case studies. Specifically, I take psychological phenomena that are thought to be goal-independent and show that, in fact, goals play an essential role in their computation. The first case study investigates automatically activated likes and dislikes, also known as implicit attitudes. Current accounts of implicit attitudes oppose the goals-first framework, as they assume that implicit attitudes can be (and often are) computed in an entirely "goal-free" manner. Disputing this assumption, I present a novel hypothesis about the computational and representational underpinnings of implicit attitudes, according to which goals are essential to the computation of all implicit attitudes. I provide support for this hypothesis with 10 experiments (Melnikoff & Bailey, 2018), and in 3 further experiments explore a surprising implication of this hypothesis. Specifically, I show that implicit attitude measures are highly sensitive to self-presentation goals because implicit attitudes themselves are highly sensitive to self-presentation goals. Accordingly, implicit attitude measures cannot be used to circumvent self-presentation, though this is precisely why they were developed, and is a major source of their sustained popularity.The second case study concerns people's beliefs about who others are "deep down". These so-called "true self beliefs" are thought to be universal and invariant. In contrast to this view, but in line with the goals-first framework, I present 7 experiments suggesting that true self beliefs are highly dynamic mental representations constructed in the service of people's current goals.Collectively, the arguments and empirical findings that I present in Chapters 1 through 7 support an account of cognition and action that places goals at its core. In Chapter 8, I consider the implications of such an account for the field of cognitive science. One implication, I suggest, is the need to re-frame one of the field's defining problems - namely, the problem of the homunculus, which refers to the fact that a "little man in the head" is often postulated as the ultimate controller of thought and action. Most attempts to "banish the homunculus" have ignored the question of how goals are caused, focusing instead on deconstructing lower-level mechanisms of control. I suggest that, in light of the plausibility of the goals-first framework, understanding how goals are caused should be at the top of the agenda of cognitive science. I conclude by attempting to lay some groundwork for answering this question, drawing from "predictive processing" models of brain function. |