[Somean] question about inferences

[E. Allyn Smith]

Hi everyone,

Do any of you have a reference for whoever did studies showing how  
inferences humans draw differ from inferences drawn on the basis of  
logic (where the following is the "classic" example I show my classes)?

Example:  I tell you a story about a woman named Heather and how she  
was involved in social justice agendas at Woodstock, etc.  Then I ask  
you to estimate how likely you think she is to be a corporate lawyer  
today.  Once you've done that, I ask you to estimate how likely you  
think she is to be a corporate lawyer *and* an activist in her spare  
time.  Logic says the second number should be smaller than the first  
since you're adding a condition to the previous one and so if the  
first is a set of all the worlds where Heather ends up a lawyer, those  
where she's a lawyer and activist are a subset of the lawyer worlds.   
Nevertheless, the vast majority of people make the second number  
bigger than the first because they think that being an activist is  
more in keeping with what you told them about Heather's past, etc.

Alternatively, do you know of anyone showing that people sometimes  
don't recognize when some statements follow from other statements even  
though they're technically an entailment relation?  I mean, I know  
that as semanticists we'd just be frustrated talking to someone who  
didn't compute entailments, but I feel like this sometimes happens in  
"the real world" when I talk to random people... (note: i'm not  
advocating giving up thinking about entailments just because some  
people ignore some of them some of the time, I'm just curious)

Thanks,
Elizabeth