Situated Simulation in the Human Conceptual System: An Overview
Concepts are not stored as abstract definitions but function as simulators that recreate sensory, motor, and emotional patterns from past experience. Thinking involves running context-specific simulations that prepare an organism to understand and act within a situation.
In his influential 2003 paper, Situated Simulation in the Human Conceptual System, cognitive scientist Lawrence W. Barsalou presents a comprehensive rethinking of how concepts are represented and used by the human mind. Rather than treating concepts as abstract, symbolic entries in a mental database, Barsalou argues that conceptual knowledge is grounded in the same sensory, motor, and affective systems that support perception and action. Concepts, on this view, are not static definitions but dynamic capacities to simulate situations.
The paper is both theoretical and empirical. Barsalou reviews competing theories of the conceptual system, introduces the framework of situated simulation, and surveys a wide range of behavioural and neuroscientific evidence supporting this account. The result is a model of cognition that emphasises embodiment, context-sensitivity, and action-readiness.
NLP trainees take note: only idiots will interpret all this as: The brain cannot tell the difference between the real thing and the concept of the thing.
The Problem of Conceptual Representation
At its core, the paper addresses a foundational question in cognitive science: what is a concept, and how is it represented in the mind? Traditional approaches assume that concepts are amodal, represented in a symbolic format distinct from perception and action. On these views, conceptual knowledge resembles an internal encyclopaedia: stable, decontextualised, and organised primarily by taxonomic similarity.
Barsalou challenges this assumption. He argues that such models fail to account for the flexibility, context-sensitivity, and action-oriented nature of human thought. Everyday cognition requires not just knowing what things are, but knowing how to interact with them in specific situations. This, he suggests, requires a fundamentally different kind of representational system.
Four Competing Theories of the Conceptual System
Barsalou begins by reviewing four broad theoretical approaches to concepts, comparing them along five dimensions: architecture, representational format, abstraction, stability, and organisation.
Semantic Memory Models
Classic semantic memory theories propose that concepts are stored as abstract representations, such as feature lists, prototypes, or semantic networks, separate from sensory and motor systems.
These representations are typically:
- Modular (distinct from perception and action)
- Amodal (symbolic rather than sensory)
- Decontextualised
- Stable across situations
- Organised taxonomically
While influential, Barsalou argues that these models struggle to explain how conceptual knowledge supports real-time interaction with the environment.
Exemplar Models
Exemplar models store memories of individual category instances rather than abstract summaries. These models capture variability and context better than semantic memory theories, but they often still assume amodal representations and modular storage. Although exemplar models are more situated, they remain relatively static and do not fully integrate perception and action.
Feed-Forward Connectionist Models
Connectionist networks represent concepts as patterns of activation across units, with learning producing statistical regularities. These models introduce dynamism and graded structure, but in their standard form they still separate perceptual input from conceptual representations, preserving an amodal and modular architecture.
Situated Simulation Theory
Barsalou’s alternative, the situated simulation theory, differs from the others on all five dimensions. It proposes that:
- The conceptual system is non-modular, sharing mechanisms with perception and action
- Representations are modal, reusing sensory-motor systems
- Concepts are contextualised, not abstract summaries
- Conceptual representations are dynamical, varying with situation and goal
- The system is organised around the action-environment interface, not taxonomy
Concepts as Simulators
A central proposal of the paper is that a concept is not a stored representation, but a simulator. A simulator is a cognitive mechanism that can generate multiple simulations of a category, depending on context, goals, and task demands.
For example, the concept CAR does not correspond to a single abstract definition. Instead, it can produce simulations of driving, parking, repairing, hearing traffic, or feeling acceleration, each involving different sensory, motor, and affective systems. These simulations are partial re-enactments of previous perceptual and action states.
On this view, thinking is not the manipulation of symbols but the controlled activation of sensory-motor systems in the absence of direct input.
Situated Conceptualisations
Barsalou introduces the idea of a situated conceptualisation: a context-specific package of inferences generated by a concept to support action in a particular situation. Each situated conceptualisation typically includes:
- Contextually relevant properties of the focal category
- Information about the background setting
- Likely actions available to the agent
- Associated internal states (emotions, evaluations, expectations)
These elements are not added post hoc. They are integral to how concepts function. To conceptualise something is, in effect, to place oneself in a simulated situation involving that thing.
Evidence for Modal and Non-Modular Representations
A substantial portion of the paper reviews empirical evidence supporting the claim that conceptual processing recruits sensory-motor systems.
Behavioural Evidence
Across domains such as perception, memory, language comprehension, and reasoning, studies show that variables like size, shape, orientation, occlusion, and sensory modality influence conceptual tasks. For example:
- People verify properties more slowly when simulated objects are larger or occluded
- Language comprehension is faster when the visual shapes implied by text match subsequently presented images
- Switching between sensory modalities during conceptual processing incurs measurable costs
These effects are difficult to explain using purely amodal representations, but follow naturally if people are running perceptual simulations.
Neuroscientific Evidence
Lesion and neuroimaging studies further support the simulation account. Damage to specific sensory-motor systems selectively impairs conceptual knowledge that depends on those systems (for example, visual knowledge for animals or motor knowledge for tools). Functional imaging studies show that processing different categories activates modality-specific brain regions associated with perception and action.
Importantly, conceptual activation typically occurs in areas adjacent to, but not identical with, primary sensory cortices, suggesting partial reenactment rather than full perception.
From Taxonomies to Action
A key implication of situated simulation theory is that the conceptual system is not primarily organised as a hierarchy of categories. Instead, it is structured around the practical problem of acting effectively in the world.
Taxonomic organisation still exists, but it plays a secondary role. More central are flexible, goal-derived groupings that bind actions to environmental affordances. Concepts support real-time prediction, coordination, and decision-making, rather than abstract classification alone.
Theoretical Implications
Barsalou addresses a common objection to grounded theories: that perceptual representations lack the expressive power needed for abstract thought. He argues that this objection is theoretical rather than empirical. In principle, systems built from simulations can support abstraction, reasoning, counterfactual thinking, and conceptual combination.
While a complete computational implementation remains a challenge, the paper provides a strong existence proof that a simulation-based conceptual system is viable.
Situated simulation theory offers a coherent and empirically grounded alternative to traditional symbolic accounts of concepts. By treating concepts as dynamic, embodied, and action-oriented capacities rather than static definitions, Barsalou reframes how meaning, understanding, and cognition are understood.
For students new to this work, the central takeaway is simple but far-reaching: to think about something is to simulate interacting with it. Conceptual knowledge is not stored apart from experience; it is constituted by structured reactivations of experience itself.
Study Movement and Cognition
Apply the science of situated simulation to clinical change work through the Metaphors of Movement model.
Situated Simulation Worksheet
Identifying the Four Pillars of a Client’s Internal Simulation
According to Barsalou, a concept is a simulator. Use this checklist to decode the simulation the client is currently running regarding their “Problem” or “Goal.”
1. Focal Category Properties
What are the specific qualities of the “thing” in the metaphor?
- Is it heavy, sharp, transparent, or opaque?
- Does it have a specific texture or temperature?
- Simulation Check: If it’s “heavy,” is the client’s body showing signs of physical strain?
2. The Background Setting
What is the “Ground” and “Medium” of the simulation?
- Where is the client standing? (Mud, path, ledge?)
- What is the weather or lighting? (Dark, foggy, blinding?)
- Simulation Check: Does the setting explain why the client is currently stationary?
3. Agentive Action Readiness
What actions does the simulation actually afford?
- Is the client positioned to move, or are they braced for impact?
- Are their hands free or occupied?
- Simulation Check: Does the current body posture match the reported “inability to act”?
4. Internal State/Affect
What expectations or evaluations are “built-in”?
- Is there a feeling of impending collapse?
- Is the “future” (forward vector) simulated as a threat?
- Simulation Check: Are the client’s micro-expressions congruent with the metaphorical terrain?
The “Idiocy” Filter:
Remember: You are not “imagining” with the client. You are observing the neurological re-enactment of their lived experience. If their simulation says there is a wall, do not tell them to imagine it gone; ask them what their hands are currently touching.
The Action-Environment Interface
Ready to deepen your clinical observation?
