Edited by: Ivan Toni, Radboud University, Netherlands
Reviewed by: Helen Tager Flusberg, University of Massachusetts, USA; Ivan Enrici, University of Turin, Italy
*Correspondence: Roel M. Willems, Helen Wills Neuroscience Institute, University of California Berkeley, 210 Barker Hall, Berkeley, CA, USA. e-mail:
This is an open-access article subject to an exclusive license agreement between the authors and the Frontiers Research Foundation, which permits unrestricted use, distribution, and reproduction in any medium, provided the original authors and source are credited.
The human capacity to communicate has been hypothesized to be causally dependent upon language. Intuitively this seems plausible since most communication relies on language. Moreover, intention recognition abilities (as a necessary prerequisite for communication) and language development seem to co-develop. Here we review evidence from neuroimaging as well as from neuropsychology to evaluate the relationship between communicative and linguistic abilities. Our review indicates that communicative abilities are best considered as neurally distinct from language abilities. This conclusion is based upon evidence showing that humans rely on different cortical systems when designing a communicative message for someone else as compared to when performing core linguistic tasks, as well as upon observations of individuals with severe language loss after extensive lesions to the language system, who are still able to perform tasks involving intention understanding.
Communication can be viewed as a matter of coding and de-coding linguistic information. The speaker codes information and puts his thoughts into words, while the listener de-codes the linguistic information, taking the input from the speaker and translating it back into a thought. In this scenario, it is the code (in this case language) that matters for communication. Individuals with a common code can communicate because they share that code. This is an intuitively appealing view given that communication in our everyday lives so often relies on language, be it in face-to-face conversation, talking on the phone, writing an e-mail, or other forms of exchange. The position that it is the code that matters for communication is nicely phrased by the philosopher John Searle:
“One can in certain special circumstances ‘request’ someone to leave the room without employing any conventions, but unless someone has a language one cannot request of someone that he e.g., undertake a research project on the problem of diagnosing and treating mononucleosis in undergraduates in American universities.” (Searle,
By this view, we are capable of communicating to some degree without language, but real communication requires language and all essential communication is linguistic. In terms of cognitive architecture, this has led to the proposal that understanding others and communicating with others by necessity involves the language system (e.g., Carruthers,
In the context of this special issue (
It is obvious that literal de-coding of linguistic utterances cannot be an adequate and complete explanation for communicative behavior (it is “commonsensical” in Sperber and Wilson's words; Sperber and Wilson, “Do you know what time it is?” “Oh man, a beer would do me good after all this hard work”
The expected answer to (1) is not just the affirmative “yes” (except in slapstick movies) and (2) may be more adequately interpreted as a request for a drink than as a factual statement about the belief of the speaker with regard to the relationship between beer and emotional wellbeing. In natural communication it is the capacity to infer someone's intention from an utterance which seems more fundamental than the linguistic de-coding of the message. This common-sense statement has been discussed and refined at length in the study of pragmatics (“the study of language usage,” Levinson,
Although it may seem trivial to claim that language and communication are not the same cognitive construct, there is a potent literature arguing that the cognitive system for language crucially underlies our ability to infer others’ intentions. Some argue that the structure of human language is crucial for representing higher-order beliefs, as required for mentalizing (Carruthers,
A second proposal is that verbs describing speech or cognitive-mentalizing states such as thinking and remembering are necessary for representing the intentions of others. An intriguing demonstration supporting such a link is the finding that deaf adults learning a sign language only start to perform well on false belief tasks after they master typical “mentalizing” verbs such as “believe” or “think” (Pyers and Senghas,
In this paper we explore the question of the relationship between language and communication/intention understanding from the perspective of human neuroscience. In particular, we address the question whether there is evidence from neuroimaging and neuropsychological studies to suggest that language and communication are supported by overlapping or distinct parts of the brain.
If inferences regarding someone's beliefs and intentions necessarily require the resources of the language faculty, either from the lexicon (e.g., mentalizing verbs) or the recursive capacity of the grammar in embedding one proposition within another, then individuals with severe language impairment would fail on theory of mind (ToM) tasks. Patients with severe agrammatic aphasia usually display deficits in the comprehension and production of verbs, and these impairments are particularly evident on low imageability, abstract verbs such as those that describe mental states (McCarthy and Warrington,
Despite the presence of such profound language impairments, patients with severe aphasia are able to succeed on tests of false belief understanding (Varley and Siegal,
With regard to the inter-relationship between language and communication, patients with severe aphasia have often been observed to communicate better than they talk. Through the use of alternative communicative resources such as drawing, facial expression, and gesture some severely aphasic individuals are able to convey quite sophisticated messages (see Siegal and Varley,
In a recent study, we investigated the capacities for communicative intention generation in several aphasic patients in an experimentally controlled set-up (Willems et al., submitted). Three profoundly aphasic patients engaged in a communicative paradigm called tacit communication game (TCG), which involves two players with different communicative roles. The paradigm consists of a 3 × 3 grid on which each player can move around his/her visual token, consisting of a simple shape (Figure
Adopting a different experimental strategy, Bara et al. (
In addition to illuminating the role of language in theory of mind reasoning, these studies indicate the continuing importance of patient-lesion studies in cognitive neuroscience. While functional imaging studies reveal the activations associated with a particular behavioral-cognitive performance, the lesion method represents a means of determining whether the activations reflect a necessary neurocognitive component of the processing network (Bird et al.,
Very few neuroimaging studies have looked directly at the relationship between communicative and linguistic abilities. There is a relatively large literature investigating the neural basis of “mentalizing” or ToM (see Amodio and Frith,
Sassa et al. (
Willems et al. (
The distinct manipulation of Communicative intent and Linguistic difficulty was neatly reflected in activation patterns in different brain regions (Figure
Both these studies suggest that producing a communicative act for another person relies on different brain areas than those involved in language. Furthermore, the areas that are activated during communicative message generation are those that have previously been observed to be activated in response to mentalizing tasks.
Besides these direct comparisons of language and communicative processes, there have a been a number of studies investigating intention understanding during communication as such, without necessarily focusing on the relationship with language. As we noted above, the neuroimaging literature on mentalizing is extensive (see Amodio and Frith,
Walter et al. (
Kampe et al. (
Finally, Noordzij et al. (
The work that we have reviewed here argues for a neural dissociation between communicative and linguistic capabilities. First, there is evidence from lesion patients who despite severe damage to the language system perform well on mentalizing tasks, as well as on tasks involving the generation of a communicative message for another person. Second, neuroimaging in the neurologically healthy population indicates that distinct parts of the brain are involved in the generation of a communicative message as compared to linguistic processes. Hence, our main conclusion is that communicative abilities should be best understood as neurally – and cognitively – distinct from language and that successful communication does not necessitate, nor rely upon a functioning language system. We take this as strong evidence for the proposal that communicative and linguistic abilities are cognitively distinct (Tomasello et al.,
It would be a mistake to interpret our conclusion of separate cognitive capacities for language and communication as meaning that language and communication have little to do with each other. On the contrary, as we described in the introduction, it is trivially the case that language is used mainly and perhaps almost exclusively in a communicative manner in everyday life. Indeed, part of the success story of the human species is due to its capacity to use language as a communicative device.
The fact that our capacity to understand the intentions of others is interlinked with normal language use, has led some to propose an “extended language network,” encompassing mentalizing related areas such as MPFC as well as areas more traditionally implicated in language (Ferstl et al.,
However, we feel that incorporating mentalizing abilities into an extended language network is not a helpful conceptualization. We showed evidence for a separation of mentalizing abilities and linguistic abilities in the human brain, such as in the case of patients who have lost the capacity for language, but are still able to communicate. Moreover, lexico-semantic processing can be distinguished from communicative message generation in the healthy human brain. The separation of linguistic and communicative abilities therefore seems a more fruitful characterization rather than calling both “language,” and allows for some forms of communication that are not linguistic. Although it is clear that there is not a single, monolithic neural network only involved in language (see e.g., Willems and Hagoort,
There is a sizeable literature which argues for the opposite conclusion to the one that we have reached, namely that language and mentalizing/communicative abilities are causally related and that the capacity to mentalize about others’ beliefs, intentions and desires crucially depends upon language abilities. An important source of evidence for this position is the observation that performance on false belief tasks correlates with several aspects of language development (Milligan et al.,
Some have argued that, although the abilities of infants show that intention recognition can precede language development, it is not until relatively late in preschool years (around 4 years of age) that children develop “the real thing” for ToM, namely false belief understanding (Pyers,
On top of the evidence from pre-verbal infants, we described findings from severely aphasic patients who, despite their severe limitation in language ability, are able to pass false belief tests. This casts further doubt upon the relationship between language and mentalizing abilities in the sense that even on standard measures of mentalizing, performance can be maintained without a fully functioning language system. It is possible that the role of language in mentalizing may be restricted to configuring the capacity for ToM in early childhood, and thus if language is impaired in later life as in acquired aphasia, mentalizing ability is not lost with the loss of language. However, the convergence of evidence from infancy and adults with aphasia strengthens the case as to the considerable autonomy between these two cognitive capacities.
In summary, we reviewed evidence from neuroimaging in healthy participants as well as results from neuropsychological populations which show that the generation of a communicative message is best thought of as a capacity which is distinct from core linguistic processes. Hence the perspective from neuroscience compellingly argues for loosening the presumed causal ties between communicative abilities and language, and between mentalizing abilities and language (e.g., Sperber and Wilson,
A similar conclusion can be reached from research in infants. The claimed interdependence of mentalizing and language seems to be mainly due to the heavy reliance in experimental studies on false belief tasks, which are theoretically well motivated (Wimmer and Perner,
Given the paucity of available data, there are many directions for future research. First, future work on healthy adults should be aimed at investigating how the two neural systems interact since they are obviously closely related. Second, research in healthy as well as in neuropsychological populations should be used to gain more adequate tools for assessment and improvement of communicative abilities in those with severe language difficulties. Finally, neuroimaging work in developing populations should investigate the intricate interplay between neural language development and development of mentalizing abilities.
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
Supported by the Netherlands Organisation for Scientific Research (NWO Rubicon 446-08-008) and the Niels Stensen Foundation. Publication costs of this article were paid through a grant from the Netherlands Organisation for Scientific Research Open Access initiative. We are most grateful to the constructive criticism and recommendations from two reviewers.