Two people with severe aphasia, a language impairment caused by brain damage, solved logic problems
Two people with severe aphasia, a language impairment caused by brain damage, solved logic problems
A study in PNAS compared their performance with fMRI data from healthy participants and found that the brain's language regions were not engaged when participants searched for rules or evaluated deductive conclusions.
On July 6, a team led by MIT neuroscientist Evelina Fedorenko published a study addressing a longstanding debate about whether words serve as an internal medium for formal reasoning. Language helps us formulate a problem, state its conditions, and explain the answer. The authors examined the internal mechanism that people use to hold these conditions in mind and derive a conclusion from them.
The functional MRI study included 29 adults. This method indirectly measures activity in brain regions through changes in blood flow. The researchers first identified each participant's language network by finding regions that responded more strongly to sentences than to strings of meaningless words. Different subgroups of the sample then searched for a rule that transformed one list of numbers into another, applied a given rule to new examples, evaluated syllogisms with “if... then...” conditions, or selected the missing shape in geometric matrices.
Comparing rule discovery with the application of a given rule allowed the researchers to distinguish the search process itself from following an instruction. The language network showed no significant response during the deductive tasks. During rule discovery, its response was more than four times weaker than its response to reading sentences. Rule discovery engaged the multiple-demand network, a distributed set of frontal and parietal regions that is active during difficult, goal-directed tasks. Complex deductive reasoning engaged a different set of frontal and parietal regions.
The two participants with profound aphasia had extensive damage to language regions in the left hemisphere and performed at approximately chance level on several grammar tests. In the numerical task, they identified 19 of 25 and 39 of 40 rules. In the geometric matrix tasks, they correctly solved 25 of 30 and 26 of 30 problems. They could communicate a rule they had identified through a gesture or drawing.
In the tasks studied, finding a general rule and evaluating a logical conclusion relied on systems outside the language network. After brain damage, language ability and the ability to infer a rule from examples should therefore be assessed separately.