Is language the key to reasoning? July 2004 - 667 words Evidence - - PowerPoint PPT Presentation

Michael Siegal

Is language the key to reasoning? Evidence from deafness, aphasia and autism

University of Sheffield, UK University of Trieste, Italy

Editorial - ‘Language, Conceptual Development, and the Essence of Human Cognition’ Trends in Cognitive Sciences, July 2004 - 667 words Relationship between language and the origin and development of concepts Became ‘Language and Conceptual Development’ ‘essence’ lost Questions: Language & spatial relations? Language & numerical computations? Language & causality? Language & the concept of belief? How do children come to represent beliefs as false? If dependent on language, when might we see these effects and in what form? But what do we mean by language? Is it in the form of the grammar of language that allows a structure that permits us to think through propositions

• r

Does this consist of conversational pragmatics - understanding the purpose and relevance of questions? The ‘Sally-Anne’ theory of mind task However, children may believe that the test question refers to where Sally has to look, or should look, for the object rather where will Sally look first (in a test of their understanding of false beliefs). ‘THEORY OF MIND’ FALSE BELIEF TASKS

(Siegal & Beattie, Cognition,1991; Yazdi, German, Defeyer, & Siegal, Cognition, 2006)

Children are told a story such as: Sam has a kitten. Sam wants to find his kitten. Sam thinks it is in the lounge. It really is in the kitchen.

• 1. Where will Sam look for his kitten? (standard

question)

• 2. Where will Sam look first for his kitten? (look first

question) RESULTS: Children aged 3 years often answer the standard question incorrectly but generally succeed on the look first question. Most 4-year-olds succeed on both questions.

On ‘standard’ ToM tasks, children may misunderstand the experimenter’s intention. They may think that the relevance and purpose of the question is to indicate where the story character should look, or must look, to find the object rather than where will the story character look first. However, according to the view that has often been expressed on the link between language and ToM, language provides a structure that “scaffolds” propositional reasoning about mental states. One specific claim is that syntax permits humans to entertain false beliefs and to reason out solutions to ToM tasks.

Two types of tests of the effects of language on reasoning: grammar vs. timing of language exposure

Deaf children who have different exposure to conversation but have similar proficiency in the grammar of their sign language Persons with RHD with impaired conversational understanding but retain grammar and persons with aphasia following damage to LH language areas

Two main groups of deaf children

• Native signing deaf
• Late signing deaf

Deaf children born to deaf parents, receive sign language input from birth Access to conversation Deaf children born to hearing parents, receive no early sign language input Little access to conversation

ToM scores of native and late signers (from Woolfe, Want, & Siegal, Child Dev., 2002)

Controlled for BSL

Senghas et al., 2004: Examples of gestures accompanying videos shown to Nicaraguan deaf and hearing Spanish speakers Senghas et al. findings challenge the notion that language evolves through cultural transmission. Signposts to the essence of language (Siegal, Science, 2004)

Same pattern of ToM findings in Nicaragua for the deaf as for the deaf tested in England, Australia, the US, France, Sweden and other countries. Younger cohorts of deaf children with earlier access to a sign language outperform the first older cohort with later access. Other areas of cognitive development (such as number) are yet to be tested.

But how about the role of grammar? What specific prediction can be made?

For example, de Villiers and de Villiers have proposed that proficiency in syntax, as shown by sentence complementation , has a decisive role in supporting ToM reasoning in that the grammar of natural language provides a code through which one proposition (false) can be embedded within another (e.g., understanding sentences such as “John thought (falsely) that the cookies were in the cupboard”). No direct test here of the specific syntax of complementation hypothesis. However, it is clear that proficiency in grammar does not guarantee success on ToM tasks for these reasons:

• 1. Normal 3-year-olds spontaneously produce

sentences that involve the syntax of complementation before they succeed on standard ToM tasks.

• 2. Moreover, children solve ‘pretense’ tasks that

require understanding of sentence complementation - tasks that have the same structure as ToM tasks except that the story character is said to pretend, rather than to think, that an object is in a (false) location.

• 3. There are some languages - both sign and spoken -

in which complementation structure does not exist: These use "clausal adjuncts" rather than clausal complements so that (a) below can't be

• used. Instead, there is (b):

(a) John told everyone that Mary washed the car. (b) Mary having washed the car, John told everyone (it).

• 4. ToM and causal reasoning is present in the

absence of grammar (Varley, Siegal, & Want, TICS, 2001) RH damage can result in impairment in ToM reasoning that is supported by conversational pragmatics and co-opted systems involving visuo- spatial processing. Yet grammar in RHD patients is spared (Surian & Siegal, Brain and Language, 2001). Is there a double dissociation? Do patients with damage to LH language areas retain proficiency

• n ToM reasoning tasks even though they have

become agrammatic? CASE STUDY OF SA (Varley & Siegal, Current Biology, 2000; Siegal & Varley, Social Neuroscience, 2006)

SA was a 50 year old right-handed male who was a

retired police sergeant. Four years prior to our study, he became suddenly aphasic. The neurological examination revealed the presence

• f a subdural empyema (a focal bacterial infection) in

the left sylvian fissure, with accompanying meningitis.

SA Scan

• Sub-dural empyema left

sylvian fissure

• Vessel wall damage left

middle cerebral artery

• Large temporal lobe

lesion, extending posterior frontal and inferior parietal zones SA is very unusual . He became an agrammatical aphasic without any explicit sign of possessing grammar in the form of propositional language. In describing pictures, SA would use word strings such as: Woman a age the red scissor A asia as man as black a shoes

However, SA can do causal reasoning

and ToM tasks.

Theory of Mind: SA

Standard format first-order false belief task (Smarties/changed container) Trained on predicates ‘think’ ‘is really’

SA’s performance

• n ToM tasks

Recognising the false belief that a book really contains a necklace Recognising the true belief that a Russian doll contains four other dolls. Recognising the true belief that a box of tissues actually contains tissues.

Another patient with aphasia, MR, also provides evidence for the double dissociation between grammar and ToM

Picture format first-order false-belief task Trained on predicates ‘think’ and ‘is really’ 10 trials on false-belief and true-belief pictures 20/20 correct on both FB/TB and reality questions

ToM in MR (Varley et al., Neurocase, 2001)

Language and the development and maintenance of ToM

Evidence from both aphasia and deafness rejects the claim that, in ToM and causal reasoning, natural language grammar instantiates thinking and reasoning. However, both SA and MR had considerable lexical knowledge still available, and exploration of the role of the lexicon versus grammar in various cognitive domains is necessary. Performance on ToM tasks is not simply a matter of executive functioning - a cluster of higher order capacities that include:

• selective attention
• behavioural planning and response inhibition,
• the manipulation of information in problem-solving.

Late signing deaf children do well on measures of EF but do poorly on ToM reasoning tasks. Following brain damage, some persons do well on ToM but do poorly on measures of EF - Aphasia: Varley et al., Neurocase, 2001 Medial frontal lobe damage: Bird et al., Brain, 2004 Syntax and executive functioning may be seen as ‘co-opted systems’ that support ToM performance in normal children and adults. Despite impairments in syntax and on EF measures, there is success on pictorial ToM tasks that do not involve syntax and on ‘look first’ tasks. In at least in the ToM case, immersion in conversation - and not simply exposure to the syntax

• f language - is a spur for development.

Conversational experience serves as a gateway to others’ beliefs. It alerts children that speakers are epistemic subjects who store and seek to provide information about the world and, in doing so, allows access to a world of referents and propositions about intangible objects, creating the potential for imagining the past and future In the case of ToM, there may be a sensitive,

• ptimal, or critical period in which some minimal

exposure to conversation is necessary to ‘trigger’ or ‘tune’ development. This position is parallel to recent findings in language acquisition more generally. “Look first” test questions do help normally developing children as well as adult patients with right hemisphere damage to succeed on ToM tasks However, “look first questions” do not help either the late signing deaf or children with autism (with possible damage to the OFC- amygdala system that may be viewed as “core” neural substrate to ToM).

If language in the form of grammar is not the basis for ToM reasoning, might it still be regarded as the basis of other cognitive domains such as mathematics? We have started to address this question with SA and other patients with severe grammatical impairments (Varley et al. (2005) Agrammatic but numerate. PNAS, 102, 3519-3524) In keeping with the results of studies indicating a dissociation between grammar and ToM, there appears to be a dissociation between grammar and numerical cognition. In both the ToM and number studies, patients had a ‘language configured’ mind in the first place. Yet the results are consistent with others that point to the independence of ToM, number and grammar in childhood as specialized cognitive faculties. What about the consequences for autism? Might there be some sort of connection between children with autism who do poorly on ToM tasks and the performance of late signing deaf children?

Is language the key to reasoning?

Is it the case that certain children with autism are impaired in a form of auditory processing, or have been impaired early in development? Does hearing impairment interfere with early joint attention and participation in conversational interaction that seems necessary for the expression of the ToM reasoning? Yes Maybe Gervais et al. (2004). Nature Neuroscience, 7, 801-2. Adult males diagnosed with autism did not show activation in voice selective STS regions in response to vocal sounds but displayed a normal activation pattern in response to nonvocal sounds. Kuhl et al. (2005). Developmental Sci., 8, F1-F12. Tested 29 children with ASD aged 32 to 52 months

• n preference for ‘motherese’ speech over non-

speech analogs and on MMN ERPs. Compared to controls, the children with ASD:

• Did not show preference for ‘motherese’ speech
• ver non-speech analogs
• Did not show a significant MNN in response to a

syllable change. If children are not exposed to an early language environment as is the case for many congenitally deaf children of hearing parents, they show later impairment in aspects of language comprehension and production. Similarly, might it be that if children do not process voices early, they show later impairment in language and conversational understanding - resulting in the presence of autism/autistic spectrum disorders accompanied by ToM difficulties?