Lots of people have reviewed this book, including these reviews by Erudito and Kim Wombles.
It inspired a whole new blog, Autism and Empathy, as a rebuttal.
And the aforementioned Kim Wombles interviewed Dr Baron-Cohen, and directed him to Rachel Cohen-Rottenberg's [Autism and Empathy site, asking for his response to her work. He, out of all the essays, research and anecdotes on that site, selected her] critique of his Empathising/Systematising theory (which is one of the foundations upon which Baron-Cohen's theory is based) [at the expense of everything else, including more recent work]. [ED: corrected details, thanks to Rachel Cohen-Rottenberg in comments on Livejournal.]
And Dr Baron-Cohen wrote a long and considered reply.
There are lots of things to say about all this, and I'm not sure I'm in a space where I can sort out and unthread the issues sufficiently to do them justice.
Some things I will say, though:
First, I think Baron-Cohen's analysis of Cognitive and Affective Empathy, while better than having one fungible "Empathy" quality, still fails to capture the whole picture. (And I think he has taken a term and jargonised what was already a specialist term.)
My model of Empathy is that there are three aspects: Cognitive (the ability to ken another's emotional state), Demonstrative (the instinctual ability to produce the appropriate or “correct” display or response), and Affect ... and this last I have to elaborate on my choice of terminology.
Affect is used in psychology to refer to emotion, but there does not usually seem to be disambiguation between the emotion as experienced, and the emotion as displayed. Terms like “Blunted” or “Labile” Affect refer to display, which may (especially for Blunted Affect) be at odds with experienced emotion. The term Affect alone, though, refers to experienced emotions as opposed to cognition or conation. If we have to invent the terminology anyway, I'm going to use Affect to the personal experience of emotion (or, in this context, empathy), and Demonstration to describe the presentation of that emotion... Blunted Affect in this model is an issue with Demonstration, as might be some types of Labile Affect where someone finds themselves crying or laughing even though they are not sad or happy. (Of course, where the lability lies in the person actually feeling inappropriately sad or happy with no reason, that goes to Affect in my model.)
OK, so we've got three aspects to Empathy: Cognitive, Affective, and Demonstrative. Or: Input, Processing, and Output.
First, one has to be able to interpret the signals coming from others, in order to determine what their internal state is. Most people don't think about this, it just happens. They are known to get it wrong, of course they are. Not even NTs are mindreaders any more than Autists are... but from our point of view, it may as well be the same thing at times. (For those "I don't always know what people are feeling either..." objections, that doesn't necessarily mean that you may as well be Autistic, or that it's proof that we're complaining about a non-problem. Most people have moments when they get it wrong. We have moments when we get it right, and it's usually the result of some conscious cognitive effort to do so. NTs see "He's happy", we see "his lips are in a smile, the corners of his eyes are crinkled, and he's not gritting his teeth, so he's probably in a good mood." NTs see "he's pretending to be happy", we see "he's smiling, but his forehead's furrowed and he's gritting his teeth. He might be happy, he might not be. Do I act as if I think he is happy, or as if he is not? What if I guess wrong?")
Then there's the processing: with the knowledge of how the other person is feeling, how does that make you feel? For Autists, after much discussion around the traps, the answer seems to be that we drown in emotion and empathy. That we feel so strongly that it overwhelms us. That may be where alexithymia starts: we feel so much that we can't get enough distance from our own feelings to be able to sort out what is what, to learn to distinguish one feeling from another. At any rate, we tend to feel strongly, and not just as a metaphor. Watching Nation's Funniest
Lastly, some of that internal state has to get out in such a way so that other people can read it. Autists tend to do badly at this, as well. We don't know how to comfort people who are sad. Slapping backs and yelling in celebration is awkward at best; threatening, disturbing and scary at worst. Our displays of emotion are not what always what others would expect, either because we don't perceive the emotional situation ("I didn't know you were upset"), because our cognition of the situation leads us to different emotional conclusions ("but that's a good thing, isn't it?"), or because we don't know how to react ‘properly’ ("um... there, there? Um...").
Baron-Cohen, in his reply to Rachel Cohen-Rottenberg, makes it clear that he disambiguates between Cognition and Demonstration, even if he conflates Affect and Demonstration. And he goes out of his way to point out what I've been saying for a while: that Autists and Psychopaths are opposites of each other.
Autists, as I describe above, are bad at Cognitive and Demonstrative Empathy -- indeed, it's one of the defining symptoms. And yet we care deeply, even if we have to ask what others take for read, or we react in strange ways. Our Affective Empathy is high. Sometimes cripplingly so.
Psychopaths, on the other hand, are notoriously charming and manipulative. That's one of their defining characteristics. They just don't care. They don't have Affective Empathy, which is why they are able to use their Cognitive and Demonstrative Empathetic skills to manipulate those around them for their amusement or profit.
And yet, the very concept of Baron-Cohen's book is to conflate the two in the public mind. It doesn't matter whether one is Autistic or Psychopathic, we are both subclasses of "Zero Degrees of Empathy", we are both damned by our characteristic deficits, and only slightly redeemed by our strength. What people take away from his book is not that Autists and Psychopaths are completely different from each other, but that we are fundamentally alike. Psychopaths are like Autists, but they're potentially dangerous. Autists are like Psychopaths, but we're the good kind. (It also, in its attribution of Zero Empathy to ‘evil’, and of Zero Empathy to Psychopaths, skips over those cases where completely neurotypical people have committed evil acts because they were able to selectively turn off their Empathy, to redefine their victims as not being of a category worthy of empathy, be they Jews, or Gay, or Poor, or Rich, or foreign, or gullible, or stupid, or, or, or... But these people were deeply empathic otherwise. Gestapo officers, as the cliché goes, loved their children. Maybe he does go into this. I haven't got a copy of his book to check.)
Even if he had used his two-variable analysis of Empathy (which, as I describe above, I already think is simplistic and hides fundamental aspects which are important for meaningful analysis), you would see a graph, not a line, where NTs are grouped in the middle (a bell curve which looks like a solid bell, not just a cross section), Autists clustered at top left (with low cognition and high affect), and Psychopaths diametrically opposite at bottom right with high cognition but low affect.
As it is, Dr Baron-Cohen's 0–6 scale is not just a gross over-simplification of a complicated collection of characteristics, it's verging on an outright lie in the way it folds reality to fit an arbitrary scale. It takes a vector (z) and extracts only the absolute scalar (|z|), and draws the conclusions from that. It's technically correct as far as it goes, but so much information is abstracted away that you can't extrapolate unless you know the rest of the story left out: it's a way of describing one part of the picture, but you can't recreate the original from the abstraction, even roughly.
And I have a problem with his data sources. He claims that his tests – such as the EQ test, and the EQ/SQ ratio – are good at distinguishing Autists from NTs. And this may well be true. But I wonder if Dr Baron-Cohen is measuring what he thinks he is measuring.
For a start, they are all full of false dichotomies, excluded middles, begged questions, and leading assumptions.
In the EQ test there are questions such as “7) Friendships and relationships are just too difficult, so I tend not to bother with them.” What if friendships and relationships are overloadingly difficult, but I make the effort anyway? Do I get marked as being more Empathic if I put myself through agonies of social interaction because I want to make the effort? Do I get marked thereby as just as Empathetic as someone who doesn't find it difficult at all?
“10) When I was a child, I enjoyed cutting up worms to see what would happen.” Again, a false assumption. What if I didn't enjoy it at all, but cut worms up anyway out of a sense of dedication to find out the answer? What if my need to know was more important than my dislike of causing injury to an animal? What if I were a psychopath, and didn't do it for enjoyment, but because I was just bored? (For the record, I never cut up worms at all if I could help it, and still to this day feel a blow in the gut if I accidentally step on a snail.)
As for “1) I can easily tell if someone else wants to enter a conversation.” That's a complete logic failure right there: If I can't tell how good I am, how can I possibly answer? What he means is “I think that I can easily tell...”, but the way the question is phrased will stop Autists up short: I can sometimes tell, but by definition I don't know when I have failed to, so I can't answer the question as asked: maybe I'm really good at it and get it right every time, maybe I get it wrong every other conversation, I don't know – I can't know.
Then there are the “People often/sometimes tell me...” questions. What if I don't interact with people: how could they tell me anything? What if they're telling me but I haven't made the connection? What if everyone is being too subtle? Or too polite to bring it up? What if they just bitch about me behind my back? Do people not tell me because it's not true, or because they're just not telling me? What's it's really asking is “I think that people think that I...”. Again, the way it's phrased will cause literal minded Autists (myself included) to have to stop and second-guess the author: If I were to answer that people don't tell me that I'm too forceful in expressing my opinions (because they don't dare to, or they are too polite to, or just can't be bothered, or whatever reason), does that put me in the same category for that question as people who can answer that people don't tell them that, because people don't think that of them?
The very questions in a test designed to test how good one is at reading minds demands that we read minds, not only those around us, but we have to read the author's mind as well, just to figure out what the questions mean.
Maybe the low scores aren't indicative of a characteristic set of behaviours, as much as they indicate that it's a test, which Autists tend to fail.
Other people have made similar critiques of his Aspie Quotient test, as featured in Wired. “I would rather go to the theater than to a museum.” ... is the museum crowded and chaotic and noisy? Is the play one in which I'm interested? Why do I have to chose? I want to do both!
Rachel Cohen-Rottenberg also made an extensive criticism of Baron-Cohen's Empathy Quotient Test (part one, two, three, and conclusion), but this was almost certainly too recent for Baron-Cohen to respond to.
She points out a lot of the things which annoy me about it, and other things besides.
For the record, here are my EQSQ results:
September 11, 2011
Respondent Average EQ Average SQ Brain Type
Males 39.0 61.2 Systemizing
Females 48.0 51.7 Empathizing
Your Score 21 109 Extreme Systemizing
What does your score mean?
Generally, the higher the score the greater your natural ability for that trait. However, the EQ test has 40 questions compared to 75 in the SQ test. As a result, although the unprocessed quotients may be used for comparing each trait ability between individuals, the absolute scores do not tell an individual if he or she has a greater tendency to empathize or systemize. A calculation taking into account the quantity of questions in each test is used to determine a person's brain type along the following continuum:Extreme Empathizing (Extreme E)
Extreme Systemizing (Extreme S)
Brain Types of Experimental Control Groups Respondent Extreme E E Balanced S Extreme S Males 0% 17% 31% 46% 6% Females 7% 47% 32% 14% 0%
The important factor to consider is not your absolute score, but the difference between the two. This indicates whether you have more natural ability as an Empathizer or a Systemizer. If your scores are about the same for your EQ and SQ, then you have well balanced empathizing-systemizing capabilities.