I couldn't fit it into this post, but there was something else I wanted to highlight in the results of this study (full text here).
It isn't all that important to the conclusions of the research itself, and it doesn't have much more than anecdotal value, since it's drawing from such a small sample, but because it runs somewhat counter to the conventional understanding of empathy in autism I wanted to showcase it anyway.
Think of it as a numerical version of Michelle Dawson's "Verbatim" series.
Anyway, here are the mean scores (totals and subscales; standard deviations shown in parentheses) of the fourteen autistic (there were fifteen, but one didn't finish all the tests) and fifteen control subjects on the Toronto Alexithymia Scale (TAS), Bermond-Vorst Alexithymia Questionnaire (BVAQ), and Interpersonal Reactivity Index (IRI)*.
Total TAS scores were significantly higher for the autistic group than for the control group, but only one subscale showed significant differences: the "difficulty describing feelings" subscale. This is not at all surprising when you consider how many autistic people --- even speaking autistic people --- even speaking autistic people who never exactly lose their capacity for speech; they're just better at it some days and worse at it other days --- say they have a lot of trouble with language, especially in the "finding the right words for whatever it is I'd like to communicate" sector.
On the BVAQ, total scores do not differ significantly between the autistic and non-autistic groups; significant differences between groups only appear on one of the five subscales --- the Insight subscale, which in other versions of the test might be called Analyzing. It reflects your ability to think about what you're feeling and why you might be feeling it. There is also a significant disparity in the whole cognitive component of the BVAQ, which is the sum of the Insight, Verbalizing and Concrete Thinking subscales. (While only the Insight subscale showed a significant difference between the autistic and non-autistic groups' average scores, the Verbalizing subscale showed a difference that, while it did not rise to statistical significance, wasn't negligible either; by contrast, scores on the Concrete Thinking subscale are virtually identical).
Silani, G., Bird, G., Brindley, R., Singer, T., Frith, C., & Frith, U. (2007). Levels of emotional awareness and autism: An fMRI study Social Neuroscience, 3 (2), 97-112 DOI: 10.1080/17470910701577020