Dan Zahavi
6 min
In contemporary cognitive science, the 'simulation theory' of mindreading—defended by figures like Alvin Goldman—suggests that we understand others by imaginatively projecting ourselves into their mental states. The core idea is that we use our own minds as a model to predict and explain the behavior of others. However, this paper argues that this approach misconstrues our most basic social interactions. By treating the mind as an 'inner' entity hidden behind a 'public' body, simulationists create an unnecessary barrier that forces us to rely on complex, inferential, or projective routines to access the minds of others.
Drawing on the work of Max Scheler, this study proposes that our primary access to other minds is not inferential or projective, but perceptual. We do not need to 'simulate' a person's anger to recognize it; we perceive the anger directly in their facial expressions and bodily gestures. Scheler argues that the mind is not a private, hidden interiority but is instead expressed through the body. Consequently, our understanding of others is rooted in a direct, non-inferential encounter with their expressive behavior. This view rejects the 'engineering' perspective of social cognition, which views social life primarily as a mechanism for predicting and controlling the actions of others.
This paper challenges the dominant 'egocentric' framework in social cognition research. By arguing that empathy is a direct form of perception rather than a simulation-plus-projection routine, the author shifts the focus from how we 'reconstruct' other minds to how we 'encounter' them. This distinction is vital for understanding human development and social interaction, as it suggests that our ability to relate to others is foundational and immediate, rather than a late-developing cognitive skill dependent on complex mental modeling.
Sam: What about the obvious objection, though? We do misread people. We do get it wrong. Doesn't the possibility of error suggest there's an inference in there somewhere that can fail?
Alex: Zahavi addresses this directly, and it's one of the more careful moves in the paper. The argument is that fallibility doesn't imply inference. Perception itself is fallible—you can misperceive a straight stick as bent in water, but that doesn't mean you were inferring the stick's shape rather than seeing it. The same logic applies here. Direct perceptual access to another's emotional state doesn't guarantee accuracy; it just means the access isn't mediated by a simulation step.
Sam: Though I'd imagine a committed simulationist could push back and say: fine, but what's doing the work in cases where we do get it right? There has to be some mechanism.
Alex: And that's probably where the paper's limits are most visible. Zahavi is working at the level of phenomenology and conceptual analysis rather than cognitive architecture. He's making a claim about the structure of the experience and the inadequacy of the simulationist framework to account for it—but he's not offering a rival computational or neural story. So the paper is stronger as a critique than as a positive alternative theory.
Sam: Which is a fair division of labor, but it does leave a gap. If direct perception of other minds is real, what's the mechanism that makes it possible?
Alex: Right, and Zahavi gestures toward embodied, enactive approaches—the idea that our bodies are already attuned to other bodies in ways that don't require central inference—but that's a research program, not a worked-out account. The paper's contribution is more deflationary: it's trying to clear the ground by showing that the problem simulation theory was invented to solve may not be the right description of the problem.
Sam: So where does that leave simulation theory? Is Zahavi saying it's completely wrong, or just overextended?
Alex: The latter, and that's an important nuance. The paper's position is that simulation might be a genuine cognitive resource for certain cases—novel situations, deliberate perspective-taking, trying to predict how someone you don't know well will respond to something unexpected. What it fails as is a default account of ordinary, face-to-face social understanding. The claim is that simulationists have taken a tool that's useful in specific circumstances and tried to make it explain the whole of social cognition.
Sam: Which is a familiar pattern in cognitive science—a mechanism that's real and important in some domain gets over-generalized until it's doing explanatory work it wasn't built for.
Alex: Exactly. And the phenomenological tradition Zahavi is drawing on has been making this kind of argument for decades—that the intellectualist picture of cognition, where everything is inference and representation, systematically misses the structure of lived experience. What this paper does is bring that critique to bear specifically on the simulation debate in a way that engages with the contemporary literature rather than just restating the classic phenomenological position.
Sam: That's a useful contribution even if it doesn't close the question. It at least clarifies what a satisfying answer would need to look like.
Alex: Which is often the most important thing a philosophical paper can do. The field has a tendency to treat simulation as the default and ask critics to disprove it. Zahavi's paper makes a serious case that the default assumption needs to be argued for, not just assumed—and that changes the shape of the debate.
Sam: Thanks for walking through that. It's a paper that rewards careful reading, and this helped me see where the real argumentative weight sits.
Alex: Thanks for listening to ResearchPod.