Reading cinnamon activates olfactory brain regions

Abstract

Some words immediately and automatically remind us of odours, smells and scents, whereas other language items do not evoke such associations. This study investigated, for the first time, the abstract linking of linguistic and odour information using modern neuroimaging techniques (functional MRI). Subjects passively read odour-related words (‘garlic’, ‘cinnamon’, ‘jasmine’) and neutral language items. The odour-related terms elicited activation in the primary olfactory cortex, which include the piriform cortex and the amygdala. Our results suggest the activation of widely distributed cortical cell assemblies in the processing of olfactory words. These distributed neuron populations extend into language areas but also reach some parts of the olfactory system. These distributed neural systems may be the basis of the processing of language elements, their related conceptual and semantic information and the associated sensory information.

Introduction

A central issue in cognitive neuroscience concerns the way in which words and their meanings are represented and processed in the brain. In the present study, we tested the hypothesis that processing words with strong olfactory associations also activates olfactory regions of the brain. The rationale behind this hypothesis was based on a theoretical perspective according to which words are processed by distributed neural assemblies with cortical topographies that reflect their meaning or, more precisely, aspects of their reference (Braitenberg and Pulvermüller, 1992, Pulvermüller, 2001, Pulvermüller, 2002, Pulvermüller, 2005). As words are frequently used together with their referent objects and actions, the cortical neurons processing word- and object-related information frequently fire together and therefore wire together, so that the information about both referent and word is bound together by cortical networks, or word webs. As referential information is processed in different parts of the cortex, action and object words would materialise as word webs with different cortical distributions.

The neurophysiological properties of such word-related cortical networks could be explained by a few neuroscientific principles (Pulvermüller, 2001). Among these, the Hebbian principle of correlation learning is especially relevant. Donald Hebb postulated that “any two cells or systems of cells that are repeatedly active at the same time will tend to become ‘associated’, so that activity in one facilitates activity in the other” (Hebb, 1949, p. 70). Therefore, if word forms frequently co-occur with non-linguistic stimuli, such as visual perceptions of objects, sounds, smells or body movements, their neuronal representations will include co-activated neurons involving specific sensory and motor information related to the referent. A consequence of this is that there are distinct neuronal assemblies for different word types, depending on the referential semantic meaning of the words (Hauk et al., 2004, Shtyrov et al., 2004, Moscoso del Prado Martin et al., in press, Pulvermüller and Hauk, in press).

Evidence for such meaning-related differential topographies was provided by neuropsychological patients and neuroimaging studies of intact brains. For example, the production or comprehension of nouns and verbs or names of animals and tools was differentially affected by brain damage (Damasio and Tranel, 1993, Daniele et al., 1994, Humphreys and Forde, 2001, Miceli et al., 1984, Miceli et al., 1988, Warrington and McCarthy, 1983, Warrington and Shallice, 1984). Positron emission tomography (PET) and functional magnetic resonance imaging (fMRI) studies have demonstrated differential activation of brain areas when action- or perceptually related words are being processed (Damasio et al., 1996, Martin et al., 1996, Martin and Chao, 2001, Moore and Price, 1999, Pulvermüller et al., 1999). More fine-grained predictions are confirmed in the case of action words that are semantically related to different parts of the body, such as face-related (e.g., ‘to lick’), hand-related (‘to pick’) or leg-related verbs (‘to kick’). Data from neurophysiological/behavioural studies (Pulvermüller et al., 2000, Pulvermüller et al., 2001), event-related fMRI (Hauk et al., 2004) and transcranial magnetic stimulation (TMS) (Pulvermüller et al., 2005a, Pulvermüller et al., 2005b) make it clear that the comprehension of these words automatically activates the motor and premotor cortex in a somatotopic manner.

PET and fMRI studies have served to identify specific brain regions that respond to olfactory stimuli. In a seminal study, Zatorre et al. (1992) demonstrated that smelling odours activated the piriform and orbitofrontal cortices. The primary olfactory cortex (POC) is located within the piriform cortex at the junction of the temporal and frontal lobes. Other neuroimaging studies have confirmed the ability of odourants to increase activity in, or near, the piriform cortex (Bengtsson et al., 2001, Cerf-Ducastel and Murphy, 2004, Dade et al., 1998, Royet et al., 2003, Small et al., 1997; but see also Sobel et al., 1998, Zald and Pardo, 1997) (see Fig. 1).

Another brain area associated with smelling is the orbitofrontal cortex (OFC), which has been identified as the secondary olfactory region. Lesions in the orbitofrontal cortex lead to deficits in discriminating odours, and this region has shown enhanced levels of activity in most neuroimaging studies of olfaction performed to date (Levy et al., 1997, Royet et al., 2001, Small et al., 1997, Yousem et al., 1997, Zatorre et al., 1992). Stronger activity is usually present in the right orbitofrontal cortex than in the homotopic area on the left, but the reverse laterality has also occasionally been reported (Royet et al., 2001, Zald and Pardo, 1997, Zald and Pardo, 2000). Olfactory-related activity has been consistently reported in the amygdala, especially during aversive stimulation (Zald and Pardo, 1997).

A hypothesis that has not been tested so far concerns the neurobiological basis of words that refer to olfactory sensations. The words ‘garlic’, ‘cinnamon’ and ‘jasmine’ are semantically linked to specific odours, and the cell assemblies processing these words in the human cortex should therefore be distributed over both language areas and olfactory regions of the brain.

We chose to test this hypothesis about the specific cortical distribution of odour word representations in a neuroimaging experiment. Specifically, we predicted that reading words whose meanings have strong olfactory associations would activate primary and secondary olfactory regions in the piriform, orbitofrontal and insular cortices and in the amygdalae more strongly than matched words with weak or absent olfactory associations. To test this, haemodynamic activity was monitored using functional MRI while subjects passively read words.