… you notice its color, the shape of the characters, its letters. Then, almost instantly, its meaning emerges: a round fruit, sweet or tart—perhaps a memory of your grandmother’s Tarte Tatin, or even the image of Newton under his tree?
Understanding a word—beyond simply seeing it—means activating an entire network of knowledge and experiences. How do these sensations and images reach our brains? Is the cognitive processing that occurs during reading a purely abstract calculation, or is the meaning of words rooted in sensory or motor experiences that resurface?
To answer these questions, we invite you to first gain an understanding of the major theoretical issues in psychology regarding access to the meaning of words; then join us as we delve into our experiments.
Individuals interact with their environment through multiple modes of input and output. A modality is a channel through which we experience the world or act upon it. It can be a sensory modality, such as vision, which allows us to perceive shapes and colors; somatosensory perception, which helps us sense touch and pain; or the verbal modality, which allows us to translate our thoughts into words.
Every time you interact with an apple, your brain goes into overdrive. If you look at it, your visual areas light up; if you take a bite, your taste centers take over… These activities are called “specific”: they are direct reflections of your senses. Then, thanks to “conductors” like the hippocampus, these fleeting sensations don’t simply vanish. They transform into lasting memory traces. Paradoxically, although these memories are built from your senses, they do not remain confined to the visual or tactile areas. They migrate to neutral “convergence zones”—true cerebral crossroads where information is stored long-term. So far, everyone more or less agrees.
Researchers disagree on what happens next: How do we reuse these memories to think or to understand what we read?
According to so-called “disembodied” theories, the brain functions like a super-powerful computer. Once an experience (eating an apple) is stored, it is transformed into an abstract symbol, a pure code. To think about the concept of an “apple,” your brain no longer needs to remember the sensation of the apple; it simply manipulates logical data, disconnected from your senses.
The theory of “embodied” cognition, on the other hand, maintains that thought remains deeply rooted in the body. According to this theory, one cannot think of an “apple” without the brain secretly activating the sensory areas associated with touch, sight, taste, and so on. Thinking, in a sense, is like reliving the physical experience at a low intensity. The meaning of a word is not an abstract code but a sensory simulation.
The debate remains open: Is our thought a sensory simulation or a purely abstract calculation? The answer may lie in the way these memory traces come to life when we read.
One possible answer may be found in studies using functional magnetic resonance imaging (fMRI).
In fact, this research shows that reading or hearing an action word such as “eat” activates the motor regions that are observed during the actual act of eating.
Furthermore, it has been shown that when an action involves the hand or foot, the corresponding motor areas are activated—there appears to be a “somatotopy,” that is, a brain organization that mirrors the topography of our body. In short: the activity triggered by sentences referring to the foot, mouth, or hand precisely overlaps with the motor areas dedicated to those same body parts.
These results therefore appear to contradict the so-called “disembodied” hypothesis, according to which somatotopic motor areas should play no role in sentence comprehension or in the representation of verb meaning.
Similarly, words related to colors engage specialized visual areas, words related to smells activate olfactory regions, and sentences describing movements engage the areas involved in the perception of movement.
However, these studies have limitations. In particular, functional MRI cannot precisely measure the moment when these activations occur. Yet the question of time is central to the debate between embodied cognition and symbolic cognition. Indeed, according to the embodied hypothesis, the activation of sensorimotor regions occurs at the very moment that meaning emerges.
To address this limitation of functional MRI, our team focused on the dynamics of brain activity in the insular cortex (results published in late 2025).
We focused specifically on this area because the insular cortex, or insula, acts as a veritable “control tower” for our sensations. It is subdivided into several subregions, organized into two major zones that work together to transform a simple nerve signal into a subjective experience. Broadly speaking, we distinguish between the anterior insula and the posterior insula. Each of these two parts processes distinct information from the body and the environment.
Thus, the posterior insula is involved in processing signals from internal organs ( interoceptive information, such as the heartbeat or visceral sensations) as well as signals related to a threat to tissue integrity ( nociceptive information, such as that generated by a cut or burn). The anterior insula, on the other hand, plays a more integrative and cognitive role, contributing to the interpretation of sensations, particularly through their emotional and cognitive dimensions.
To assess the dynamics of brain activity in the insula, we used a method that offers one of the highest levels of temporal and spatial resolution available in humans:intracranial electroencephalography (iEEG). iEEG is an invasive neuroimaging method used exclusively for medical purposes, involving the monitoring of brain electrical activity using electrodes implanted within the brain. This method thus allows for highly precise measurement of neuronal activity in terms of both location and temporal course. It is used in particular in patients with drug-resistant epilepsy (whose seizures cannot be controlled by medication) to locate the area of the brain responsible for the seizures.
We therefore collaborated with 16 patients who had been implanted with iEEGs at the Grenoble-Alpes University Hospital for medical reasons. We asked them to read short sentences belonging to different semantic categories: abstract (e.g., “I am thinking”), action-related (e.g., “I am running”), and somatosensory (e.g., “I am burning myself”).
Our results show thatthe posterior insula is not only involved in the perception of sensations or pain that are actually physically induced, but can also be selectively activated during the processing of sentences describing somatosensory sensations, such as pain (for example, “I’m burning myself”). This neural response occurs very early, approximately 150 milliseconds after the word “burn” appears on the screen. Interestingly, the anterior insula does not exhibit this temporal pattern.
This extremely short time course suggests that the posterior insula may contribute to the lexical-semantic processing of sentences related to bodily sensations. In other words, when processing sentences related to sensations, the brain may simulate sensory and painful experiences, partially reactivating regions that are typically involved in the processing of actual sensations.
The involvement of the posterior insula in response to such sentences offers insights into the debate over embodied versus disembodied cognition, according to which accessing the meaning of words associated with sensations involves a partial re-experiencing of the corresponding sensory state.
In summary, there is a body of research—employing various methodologies—that challenges the so-called “disembodied” hypothesis, which dominated language theories until the end of the 20th century. The modality-specific areas of the brain—including the visual, auditory, and motor cortices—which have long been known to play a crucial role in the perception and production of word form, now also appear to be involved in the instantiation of word meaning.
Of course, many questions remain unanswered. One crucial question is undoubtedly this: How can simulations specific to a sensorimotor modality represent abstract concepts such as time, justice, or happiness? Concrete objects (such as an apple) or concrete actions (such as throwing) can be represented by sensory or motor simulations. But how can we represent ideas that we can neither perceive with our senses nor manipulate with our muscles?
The LAMI project (ANR-22-CE28-0026) is supported by the French National Research Agency (ANR), which funds project-based research in France. The ANR’s mission is to support and promote the development of basic and applied research across all disciplines and to strengthen the dialogue between science and society. For more information, visit theANR website..
This article is republished from The Conversation under a Creative Commons license. Readthe original article.
