First of all, I don't mean to insult you or your expertise on the subject. I know very little about autism spectrum disorders and nothing about this "facilitated communication" debate.
However, that Daniel's parents are unconsciously inventing his written voice and, therefore, inventing virtually everything we know about his cognition seems to be a huge assertion to make with very little evidence. The article says the parents would "hold his hand while he spells" (emphasis mine). I could see parental bias (a la ideomotor effect) playing a role in guiding the subject when the answer is one choice out of several discrete options, but it seems impossible for that kind of placebo effect to play out in spelling words, where the sequence of letters would have no meaning to the parents until a word or morpheme was already complete or almost complete, let alone sentences. Has research indicated that facilitated spelling can and, as you suggested, normally does just result in the parent talking? I really hope the answer is no...
It is certain that in the vast majority of cases, the parent/facilitator is entirely imagining the input from the subject. For instance, experiments have been done where a 'facilitated' subject who was apparently capable of complex conversations through facilitation was shown a simple object (eg an apple) where the facilitator could not see it, and then the subject was unable to name it through the facilitated communication. The APA and most psych bodies pretty much denounced it as bunk in the mid 90's (see http://www.apa.org/research/action/facilitated.aspx). Studies that find it 'works' tend to avoid lab settings and 'overly controlled settings', similar to the studies of working telepathy.
While it is not impossible for any given instance of facilitated communication to be actual communication, it is irresponsible to believe so without some pretty solid verification given that even proponents of the technique admit that in 'many' cases the facilitator was doing the communication. I believe that the most recent APA stance was that facilitated communication should be actively avoided.
The classic example of the ideomotor effect is the ouija board in which you spell out words letter by letter by "resting" your hand on a wooden planchette and easily believe that it is not you moving it. Holding someone's hand who you desperately want to communicate is only going to make that mental trap stronger.
Makes sense. I guess I never associated the ideomotor effect with spelling, and I still don't see how that would work, but I guess it does. Do you happen to have a background in magic/mentalism?
People doing this kind of facilitated communication, had previously written entire sentences, paragraphs, poems, and so on this way. And it was all, as far as the experiments show, the facilitator, not the patient, who was really writing it. Some of the facilitators had no idea; they really thought that it was the patient writing everything, and they were just helping support the patient's hand. A few of them renounced FC after this, realizing it was just them subconsciously writing for the patent, but some continued to persist in their belief that it really was the patient writing despite the strong evidence to the contrary.
Have you never had an experience where you attributed reasoning to a pet's actions? Sometimes we can tell what a pet wants, and so we can reason about what they are doing; sometimes we might be wrong about it. There are plenty of cases where you have a good guess based on other information, like the pet getting hungry at the appropriate time. If you tried doing "facilitated communication" with your pet, and really believed that it worked, you would probably wind up spelling out things like "I'm hungry" when it was hungry; and then take that as evidence that your pet really was writing that, not you.
In fact, this kind of reasoning is used all the time in research that shows that you can teach animals human language, or sign language, or the like. You get researchers who learn a lot about the animal; learn about its attitudes and ways of non-verbally expressing itself. Then they start reading things into the signs or sounds the animal makes, and present it as evidence that they've taught the animal language. But if you do any kind of experiment that takes that particular researcher out of the picture, or provides a reasonable level of blinding, all of that ability goes away, and you hear them talk about how "well, they can't perform under these kinds of stressful situations."
You might think "well, so what, it's at worst something harmless that makes the family feel better." But there have been cases where via facilitated communication, kids (or based on the above research, their facilitators), have accused their parents of sexual assault. Fathers have had to move out of their houses and fight long legal battles on the basis of an accusation that, most likely, was simply made up by prompting of the facilitator. Believing that facilitated communication works without strong evidence that it's the patient and not the facilitator in control can have some serious consequences.
But if you point this out, especially in relation to a feel-good FC story like this one about someone completing college level coursework through FC, you just sound mean and contrary. I would love to believe that advancing technology and therapeutic methods allow people who cannot learn via a conventional methods to have access to knowledge they never would have otherwise. But given the lack of strong evidence that FC really works, these feel-good stories sound more like people just trying to sell more snake-oil than an actual breakthrough.
However, that Daniel's parents are unconsciously inventing his written voice and, therefore, inventing virtually everything we know about his cognition seems to be a huge assertion to make with very little evidence. The article says the parents would "hold his hand while he spells" (emphasis mine). I could see parental bias (a la ideomotor effect) playing a role in guiding the subject when the answer is one choice out of several discrete options, but it seems impossible for that kind of placebo effect to play out in spelling words, where the sequence of letters would have no meaning to the parents until a word or morpheme was already complete or almost complete, let alone sentences. Has research indicated that facilitated spelling can and, as you suggested, normally does just result in the parent talking? I really hope the answer is no...