That's called the "framing effect," and it's usually tested by proposing a hypothetical choice to groups of volunteers in different terms: the same choice described in terms of what they could gain or what they could lose. When the choice is "framed" in terms of potential gains, people tend to take more risks than people faced with the same choice framed in terms of potential losses.
However, not everybody ultimately responds in the same way to the framing effect. This 2006 study (summarized in the above-linked BBC article) by Caltech neuroscientist Benedetto de Martino had 20 people (all college students or college graduates) make a choice, given some hypothetical sum of money, to either keep a portion of it or to "gamble," with a 40% chance of keeping the whole sum and a 60% chance of losing it all, while inside an fMRI machine. While there was a definite overall framing effect --- the subjects for whom the choice was framed as a gain (i.e., "out of $50*, you keep $20, or take a chance to win, or lose, all of it") tended to be more conservative, opting for the smaller, surer payout, while those given the opposite spiel ("out of $50, you can either lose $30 or take the chance") --- not everybody ended up making the choice suggested to them by the framing.
So far, so predictable, right? Most people are susceptible to the framing effect, but individuals differ in their degree of susceptibility, some to the effect of not being swayed at all. That's hardly earthshattering news.
No, the interesting tidbit to come out of that study is this piece of fMRI data: amygdala activity did not differ between those people susceptible to the framing effect and those who are not. What distinguished those two groups was, instead, activity in the medial and orbital prefrontal cortex, where higher reasoning and suppression of impulses occur. In anatomical terms, that means everybody, regardless of how "rational" their ultimate choice was, reacted emotionally to the choice; the people who resisted the framing successfully were just better at second-guessing their emotional reactions with calculations.
And now, finally, the thing I really wanted to talk about: in this study published in last Wednesday's Journal of Neuroscience (summarized here on Science Daily), de Martino and his colleagues posed the same scenario (i.e., the decision to keep a portion of the money or gamble for all or nothing) to a group of fourteen people on the autism spectrum and fifteen age-, sex- and IQ-matched NT controls. However, instead of using fMRI to track the subjects' brain activity during the choice, de Martino et al measured skin conductance, which is a measure of physiological arousal (i.e., anxiety). I was not as impressed with that choice of metric, as skin conductance seems to be a much less direct measure that gives us less insight into what's going on in the brain.
Quibbling aside, what they found was that the autistic group a) showed no significant framing effect and b) showed no difference in skin-conductance response to different frames. (The control group had lower skin-conductance measurements for the "gain" frame than for the "loss" frame, indicating a lower level of anxiety when the choice was phrased as one between gains rather than losses). The autistic group did show a higher overall level of skin conductance, which the study's authors attribute to greater anxiety levels in general. The inference they draw from these findings is
...that ASD subjects, although having an absolute emotional response, crucially fail to differentially engage (sic -- split infinitive!) emotional processes in response to the framing manipulation. (Italics and emphasis mine)
I thought I would pre-empt any objection that this research might be used to justify the "emotionless robot" stereotype of autistic people. Nowhere does this article deny that we experience emotions --- indeed, the autistic group was the group with the higher recorded emotional responses! It's just that our emotions aren't as readily influenced by external cues, and maybe also that our emotions are not as closely involved in our decision-making process.
De Martino leans more heavily on this latter hypothesis than I do --- I think only the former is demonstrated by this study.*** Here, he enlists his own earlier study, and some other studies of the neuroanatomy of autism, to try to bolster this point:
Interestingly, I took the findings from de Martino's 2006 study as indications that autistic people aren't necessarily detached from their emotions during decision-making. If you scroll back up and look at the results of that study, you'll find that all subjects, regardless of susceptibility to framing effect, had the same level of amygdala activation. The difference was in the prefrontal cortex. At minimum, I'd want to see the 2008 study replicated with fMRI rather than skin-conductance, so that investigators could look at the roles played by both of these brain regions in autistic decision-making.
In a previous fMRI study (De Martino et al., 2006; Kahneman and Frederick, 2007), we showed that the engagement of an amygdala-based emotional system played a key role in underpinning a framing effect. These previous results, combined with the [skin-conductance] data shown here, suggest that the failure to assign emotional salience to contextual cues and the consequential lack of behavioral bias in ASD may result from an amygdala based mechanism. A wealth of empirical data supports this hypothesis. First, histopathological abnormalities of the amygdala such as increased cell density and a reduced dendritic arborization have been described in autism (Bauman and Kemper, 1994). Moreover, Howard et al. (2000) suggested that persons with high-functioning autism showed a similar neuropsychological profile to that seen in patients with amygdala lesions ... Last, several imaging studies demonstrate blunted activation of the amygdala in ASD during tasks which involved processing of facial expressions (Baron-Cohen et al., 2000; Critchley et al., 2000).
That said, this is exactly the kind of autism research I like to see. It's descriptive, and highlights the differences between NT and autistic cognitive styles without assuming that autistics simply aren't thinking.
*In the original study, the amounts are in British pounds. I didn't bother to convert because it's a hypothetical scenario, and will work with almost any** arbitrary sum of money.
**An exception to this rule would be really large amounts of money. This game has been played with huge hypothetical sums of money, like $1 million for the sure payout and $2.5 million for the gamble. The ratio between those sums is the same as the $20 and $50 mentioned above, but more people choose to keep the sure payout in the high-value game because both numbers are so high. With those stakes, people would rather play it safe and get rich for sure than take a less-than-favorable chance to get even richer.
***You could make an argument that the lack of a risk-aversion corresponding in significance to the skin-conductance readings in the autistic group argues for a disconnect between emotions and decision-making in autism, but I don't think there's enough of a basis here to proclaim that as fact.
De Martino, B. (2006). Frames, Biases, and Rational Decision-Making in the Human Brain Science, 313 (5787), 684-687 DOI: 10.1126/science.1128356
De Martino, B., Harrison, N., Knafo, S., Bird, G., & Dolan, R. (2008). Explaining Enhanced Logical Consistency during Decision Making in Autism Journal of Neuroscience, 28 (42), 10746-10750 DOI: 10.1523/JNEUROSCI.2895-08.2008