Sunday, February 28, 2010

Skinner is not somebody who wants to be balder than a Skin Head

Yesterday, I received the visit of a friend. PhD at Stanford in one science or another, journalist, he knows about everything.

He didn't know who Skinner was. Hadn't heard the name. Even more disturbing: he knows about much more minor figures in Psychology, like Freud, who is only #2 in the list of the ten most influent psychologists on

Skinner, who wrote about superstition in the pigeon! I was mad.

I bought Walden Two, a novel that described a utopia. Said Skinner published it in 1948. According to the Wikipedia entry about the book:
When the members find a problem in their community they may design and experimentally test a possible solution, carefully documenting the results of their experiment in accordance with thescientific method.[11] If the results of their testing indicates that the proposed solution would be an improvement over their current cultural practices then they may make that experimentally validated improvement into a component of their community's culture. This cultural optimization process is called "cultural engineering."[12]
You can imagine how much I am looking forward to this book. The current edition has a preface called "Walden Two Revisited." About twenty-five years after its publication, Skinner writes about the relevance of his novel. He sure sounds Communist, and I don't mean that in a bad way.

Thursday, February 25, 2010

A toaster that toasts on both sides


Karen Pryor starts her book 'Reaching the animal mind' with this observation: that Konrad Lorenz created ethology and B. F. Skinner created behaviorism, and animal clicker training inherits from both: you need to know the nature of your partner. You have to understand the machinery.

Excellent development tools are now available for C++, Java: you can see the source code as it is being executed, dump the value of critical variables, and repeat the same code many times, stopping it at appropriate places.

Nothing of the sort exists for Dario. I bought "Cry Translator" for my iPhone. I record the Dario's cry, and it tells me if he is hungry, bored, stressed, sleepy or has discomfort. Usually, I get two or three different answers in the same number of consecutive recordings. One could qualify this app as useless, or at least useless for an infant that age, or useless for an infant with a French accent (the app comes from Spain). But it still serves a purpose: it reminds us that there are not so many types of things that can go wrong with the machinery.

But as far as debugging tools go, we don't have much else at home. We observe Dario swaying his head left and right, opening his mouth wildly, closing it on his fists, the blanket. And that, we learn, may mean that he is hungry, has gas, is tired, and, maybe, that he is stressed.

Contemplative on his mat, Dario is becoming fussy. Swaying his head left to right, opening his mouth wildly, closing it on his fists. We know what this may mean: anything. The key to successful analysis: change one variable at a time. We burp him and put him back on the mat. Same. We change his diaper and put him back on the mat. Same. We give up.

Wednesday, February 24, 2010

Return of the lama

In a recent post, I describe the lama training technique. (The word "recent" is redundant as I don't have any non-recent post relative to this one.)

There is a reason for that: it inspired me to try a technique to put Dario to sleep. I tried it when he was six-days old. It worked beautifully.

Letting the kid cry (as in controlled crying) for a while before coming back is like allowing the lama to run away, withdrawing only after a while: a big no-no according to animal training theory, much slower than the recommended approach. For instance, using that approach, a student at the EATM was able to put a cotton swab in a koala's ear in less than a year.

Hence the translation of this technique to an infant: put in bed, give finger to suck for ten second, remove. Wait for about ten seconds, give finger again, remove. Wait a little longer, give finger just before your beloved starts wiggling, and so on. At some point, rubbing your finger against his lips doesn't trigger any reaction anymore. He closes his eyes. Success.

And with a bit of luck, each night, it will be a little longer delay between a finger suck and the return of fussiness.

As I wrote: it worked beautifully.

Further attempts in the days that followed showed that the correlation was serendipitous. The only thing that I achieved is: avoid the cries while Dario is falling asleep, but I do not control the timing.

Controlled crying

The Scott Brown of sleep training techniques is called "controlled crying". You let your kid cry, but not as much as the colonel plans to do after he locks the steel door of his heart that happens to also be the door to his son's bedroom. (He swallows the key.)

Controlled crying is described in four steps by -- a name that certainly inspires trust. From the web site, I suspect it is a TV show.

Steps 1 to 3 are simple: bath, PJs, story time, cuddle, crib. But, Attention, attention: no wild hand gestures. You will be done with the first three steps in twenty minutes. Step 4 shouldn't take more than an hour. But don't worry: you have to do it only if your baby cries. I quote:

If your baby cries when you place her in her crib, start the controlled crying technique:

  • Wait five minutes before going back to the room. Make your check brief and any interaction minimal. Don’t touch, pick her up, or cuddle her – simply say, “Mommy/Daddy is here, go to sleep” and then leave, even if he is still crying.
  • If she’s still crying after 10 minutes, repeat the procedure.
  • If she’s still crying after 15 minutes repeat the procedure and continue to repeat it every 15 minutes until she’s asleep.
  • If she starts to quieten down, wait to see if she’s starting to settle to sleep – if you go in at this point you may disturb her. If she starts to cry loudly again, start your checking again.
[...] If you apply the technique properly your child shouldn’t cry for much more than an hour.
It seems very close to the "cry-it-out" solution of the colonel. Super nanny adds this insightful comment:
The key to this technique isn’t to stroke, pat or re-position your baby. This type of contact could be seen as a reward for crying and rather than reduce the crying it could teach your child to cry for a set period before you go into her and cuddle or stroke her… thus inadvertently encouraging more crying.
I dislike this technique on theoretical grounds: how is it that 'stroking, patting and re-positioning' will encourage more crying, while a warm 'daddy is here' will not?

Yet, there seems to be a strong body of evidence that this technique works within a few days. If the theory is in conflict with experimental results, the theory has to go.

This technique is not recommended before the age of six months. We hope that Dario will have been trained using other techniques by then. We have five months and two weeks left. Super Nanny says:
we feel a gradual retreat technique is more appropriate for younger babies.
This sounds appealing, but Super Nanny doesn't describe it. I find a good description in It looks great.

Sunday, February 21, 2010

On sleep training and Modus Tollens

There are different techniques to train your baby to sleep. According to, the full range goes from, at the extreme left, attachment parenting to, at the extreme right, the cry-it-out method. The left-to-right ordering has the right feel to it. I can see leftist hippies co-sleeping with their child as easily as I can see a conservative army colonel letting his kid cry to exhaustion. You will become a man, my son.

A term that recurs is that of "sleep association". For instance, the child who nurses and immediately falls asleep may be learning that nursing leads to sleep. Like most high schools students, he may also apply the Modus Tollens the wrong way, and conclude -- incorrectly -- that no nursing implies no sleep. I have difficulties to believe that an infant shall be so precocious as to be acquainted with logic, and shall at the same time make such a basic mistake in its application.

Moreover, what of Pavlov's dog? He learns that when the bell rings, food will come. To be BC (behavioristically correct), I should say: when the bell rings, he salivates. I haven't read anywhere about the more surprising result, that when food arrives without the chime, the dog doesn't salivate.

Is the dog smarter than the infant, knowing that if P implies Q, then it is not necessarily true that Not P implies Not Q? Or is he more stupid, not able to even grasp the notion of logical implication but in a feeble neuronal loop that doesn't go through the heights of the frontal lobe?

Friday, February 19, 2010

Just fussy enough

Dario is just fussy enough for me to try a variation of the lama training method.

The lama training experiment

Karen Pryor describes how to train a lama to accept to be touched. You start approaching from far away. If you approach too much, the lama will run away. So you approach just enough, until you see the lama show signs of tenseness. At that point, you give him a reward for having left you come so close: you withdraw. He relaxes. You iterate, sometimes coming a little closer, sometimes staying a little longer.

The success of this technique leads to the following hypothesis: lamas don't have the cognitive abilities to discretize continuous quantities. 100 meters and 98.3 meters? It's really roughly the same. 98.3 and 95.2? Same. And so on.

How to test this hypothesis? One could discretize the continuum for him: pre-install a set of very different landmarks (from the lama's point of view: his favorite foods?) at the stations that you will be posted.

Before we get into technicalities regarding the convergence of Cauchy sequences, let's re-center onto what's important: that it works. And the question that is on everybody's lips: can this be applied to infants.

Thursday, February 18, 2010

Animal training on babies?

Imagine my disappointment when I typed the Google query animal training on babies: there is nothing better than discussions on baby dogs and on dog meet baby.

On the other hand, clicker training on babies: we are sent to Karen Pryor's web site, with a blog entry by chipper28, that starts so:

Clicker training a baby. I’ve already discovered that these four words elicit a strikingly wide set of responses. “How could you even say that?”, “Huh?”, “Do you really think a baby’s smart enough to do that?”, “Interesting idea,” and “You can not do that to my grandson!” are a few examples. Leaving out the emotional responses for the moment, we seem to be left with either explaining what clicker training is or arguing whether our son might be approaching our dog in intellect. I haven’t actually spoken to anyone who had any real constructive ideas unfortunately.
Said chipper28 is unlikely to be Karen Pryor herself. She or he continues on her post dated 01/02/2008: "we have a four-month-old son and a four-year-old Irish Terrier."

Looks like I will get all I want. Alas, chipper28 got too busy with the training itself: this is the first and the last entry in his or her blog hosted on Karen Pryor's web site.

The visit to Pryor's site was an occasion to learn that she published a new book in 2009: Reaching the animal mind. As I have bought "Don't shoot the dog" about 15 times to give it to friends, it is proportionally not much more to buy this book without reading any review.

And so it starts

Having a kid one week of age is the perfect occasion to also give birth to a blog. For one thing: Google's generous paternal leave policy gives me seven weeks to idly muse during the times Dario is not to be attended to.

The title of the blog is directly linked to the location I am starting it: I am on my dining room table, facing the kitchen -- "une cuisine à l'américaine". On the counter top is a toaster.

I was about to call the blog "don't shoot the kid", a reference to Karen Pryor's famous "Don't shoot the dog", a giveaway on my main interest today: how to apply animal training techniques to babies.

Yet, today's interests are not tomorrow's. When Dario turns forty, the meaning of "toasted on both sides" will be as crisp and current as it is today: it means nothing. Until then, Elisa and I have acquired this infant that we are ready to be trained by.