How does the brain integrate different senses to represent the environment?
Claire Martin, researcher at CNRS-Université Paris Cité, and Boris Gurevitch, researcher at CNRS, Institut de l’Institut de l’Audition-Pasteur.
We constantly receive a stream of information from our surroundings, which we access through our senses (hearing, touch, sight, smell, taste). Our brains are challenged to combine this information to obtain a coherent perception of the world around us. For a long time, research in neurobiology has considered the different senses in a compartmentalized manner, but the study of multisensory interactions is now evolving.
Perception and illusions
We talk about multisensory interactions when information from one of the senses modifies the processing of other sensory modalities. This complex mechanism allows the brain to improve its performance in certain situations. For example, it comes into play in noisy environments where it is easier to understand the speaker by visualizing the movement of the lips. In this case, called the “cocktail effect,” the brain more easily identifies speech that matches the rhythm and shape of the lip movements that produce it. Conversely, if the sound does not match the image, it can lead to misperception.
The most striking example of this is the McGurk effect: if shown a video of a person articulating the syllable “ba” with a sound matching the syllable “ga”, the majority of people perceive the syllable “da”. The dissociation between the signals coming from the two sensory channels “sound” and “image” prevents the brain from correctly interpreting the signal and ultimately creates an illusion. As these examples show, the integration of different sensory information from the brain determines the final perception.
Sounds and smells respond to each other
The taste of food is an example of multisensory interaction due to the close relationship between three senses – smell, taste and even touch, the latter being responsible for tickling the bubbles of sparkling water, for example. Apart from this particular case, multisensory associations involving odours are poorly understood. However, they are powerful. A few years ago, researchers asked enology students at the University of Bordeaux to describe white wine artificially colored red [1].
The presence of misleading visual cues prompted the students to describe the drink’s smells using mostly red wine terms. Perception guided by the visual cues again created a perceptual illusion. In another example, participants were asked to describe their impression of crunching chips in the presence of sounds of varying frequency and intensity [2].
Surprisingly, the crunchiness and freshness of the potato chips were perceived differently depending on the associated sound. In fact, the absence of the characteristic sound of potato chips in the mouth made us doubt their quality.
Although sometimes misleading, multisensory interactions are essential. However, the brain mechanisms that make them possible still represent many unknowns.
How does our brain integrate the different senses?
In the brain, information processing is distributed among different areas that receive privileged sensory inputs. Thus, visual areas do not receive the same information as olfactory areas, and these signals are processed differently in each area. In recent years, the existence of a boundary between monosensory and multisensory areas has been questioned. Areas that until recently were thought to be specific to one sensory modality are in fact influenced by other senses. For example, in rats, the response of neurons in the auditory cortex to the distress calls of their offspring increases in the presence of the rat’s odor. This remarkable mechanism would enhance the mother’s attention to her offspring to protect it more effectively [3].
These recent findings call into question the notion of a sensory cortex specific to a sense. Information detected by our sensory sensors could actually be influenced by the first stages of processing by the nervous system. In this way, the brain could sort the information more efficiently.
Therefore, there are two hypotheses: one considers independent processing of sensory information before it converges to other areas, and the other suggests an influence of sensory information during its processing by the first stages of brain representation. To make progress in understanding these mechanisms, it is essential to determine in which brain areas and at which point in processing sensory information interacts.
However, there are still few studies of neural activity in several types of structures in response to two sensory modalities. For this reason, we trained rats to learn the association between sound and odor [4] when sound or odor is presented alone, the rat can go and get a small sugar cube as a reward. In contrast, simultaneous presentation of sound and odor carried no reward. To minimize the animals’ effort, they were gradually taught (over several days) to stop moving in the presence of sound and odor. Throughout the training process, brain waves (see inset) were recorded in various brain structures, including the olfactory bulb, primary auditory cortex, and piriform cortex. The olfactory bulb and the primary auditory cortex are so-called primary areas, as they are the first relays of the cerebral cortex where information from receptors is processed. The piriform cortex is a brain area that is primarily olfactory, but is known to process sensory information from other modalities, such as sound.
The results of this study showed a simultaneous increase in certain brain waves in olfactory areas in response to odor and sound after the rats had memorized the stimulus-reward associations. Brain waves represent the activity of neurons in the recorded structure; they would also represent a means of establishing a dialogue between nearby and distant brain areas. Therefore, this increase in simultaneous activity could reflect the establishment of an entire brain network involved in the resolution of multisensory interaction. This process would be similar to memory or attention processes, for example, which also involve networks of several interconnected areas.
More surprisingly, this work has shown that sound presented alone activates the olfactory bulb and pyriform cortex, which are primarily olfactory areas. Thus, after training, the olfactory areas behaved as if they had learned to recognize sounds. These results, as well as those of other researchers, call into question the rather hierarchical view of the brain, according to which only certain regions would integrate information from different senses. On the contrary, it seems that new brain networks can form rapidly based on the information exchanged between sensory areas to achieve the most efficient unified perception possible. However, the specificity of brain areas should not be completely dismissed.
For example, again in our study, unlike the olfactory bulb in the presence of sound, the waves recorded in the primary auditory cortex were unchanged in the presence of odor. It is possible that activity and brain waves in this brain area are less influenced by other sensory modalities, but this question remains open.
Multisensory interactions and memory: interrelated processes
The observation of responses to sounds in olfactory areas suggests that multisensory interactions contribute directly to memory mechanisms. Memory is often recovered from its fragments. An element of a scene can instantly bring us back to a completely different context from the past. This is the famous example of Proust’s Madeleine, but we can also imagine a moment of nostalgia sweeping over us as we listen to music associated with a joyful or sad moment from our past.
In the study described above, the presentation of a single sound would be able to reactivate all the information stored in memory, involving both olfactory and auditory areas. In the present research, brain waves seem to be an effective way of connecting all the areas involved in the creation or recall of a memory.
These results reinforce the idea that the brain functions as a set of interconnected networks, and the neurobiological mechanisms underlying multisensory integration are elements that allow us to understand this functioning. Moreover, in the context of pathologies or the loss of a given sensation, this knowledge will be necessary to optimize sensory rehabilitation devices.
What are brain waves?
Brain activity triggered, for example, by a sensation will activate a particular group of neurons called a neuronal assembly. These neurons may belong to the same brain area or to several areas. Within these ensembles, neurons influence each other by activating or inhibiting other neurons to which they are connected. These antagonistic actions generate rhythms that can be studied by measuring the electrical current in the vicinity of the neurons.
In humans, this current is recorded by measuring an electroencephalogram (Hans Berger, 1924). In the treatment of some pathologies it can be recorded directly in contact with the brain. The activity is most often presented as a mixture of waves of different frequencies, i.e. oscillating more or less rapidly. The first brain waves to be characterized about a hundred years ago were those occurring during relaxation, but since then many families of waves, classified according to their frequency and named with Greek letters (alpha, beta, delta, gamma, theta), have been associated with different states of wakefulness, sensory, cognitive processes, etc. These waves are now relatively well characterized and are used in brain-machine interfaces, as for example in the operation of some prosthetic limbs.