I'm struck by the notion of directionality (this may not be the best word here) of the conveyance of "truth" here.  If the harm of the fat person's inaccurate perception or belief about his own status is negatively affecting only him (he routinely gets turned down by attractive others whom he asks on dates or he can't walk from the sofa to the refrigerator without terrible pain), should the falseness of his perceptions be pointed out to him "for his own good"?  

Doing so seems to lack a basic sense of intellectual humility.  If, as has been stipulated, rationality is probabilistic (based on data/evidence/experience available to date) rather than definitive, it could be wrong (i.e., untrue).  

It seems to me, that they would need to notice the word “and.” They would need to be wary of it—not just wary, but leap back from it.

Isn't this a problem of conflicting statistical union with and intersection?  The "and" or "or" seems to suggest so.  Why would we be programmed to do this?

A bias is an obstacle to our goal of obtaining truth, and thus in our way.

In this instance, does "truth" (which many of my academic colleagues believe to be a term that is "evil and Just Not [Used]" refer to some basis in concrete reality with which an understanding of and effective/efficient ways of working with are our interest/goal?  Or can it refer to some reality (truth) that is greater than we are capable of seeing or measuring directly?  My initial sense is that it must be the former because it is in this realm that we can, in fact, use data to make decisions about probability.  But as I think about it, I suppose it might be possible, if non-physical truth/truths existed AND that they manifested somehow in the physically observable world, that we might still be able to gradually refine our thinking and behavior based on probabilities of particular things in the physical world.  Maybe?

But this seems to border on attributing non-observable causes to experienced phenomena, which seems uncomfortably close to religion.

it might be necessary to distinguish between experience and expertise, with expertise meaning “the development of a schematic principle that involves conceptual understanding of the problem,”

It seems that this same fundamental concept may be relevant to, or perhaps parallel to, a distinction between knowledge and wisdom, or information and wisdom.  The former is a collection of data typically drawn from experience, the latter is based upon the rational consideration and testing of the knowledge/information in ways that have refined it.