Caturday felid: “Methinks it is like a catcerto”

March 15, 2014 • 5:41 am

Today we have a rare guest Caturday felid; I can’t remember one since I started this site five years ago (has it really been that long?). So here’s Greg’s contribution, which shows a concerto (“Catcerto”) composed by Mindaugas Piecaitis to embellish and complement the playing of Nora the famous piano-playing cat. (The score for “Catcerto” can be found here.)  Greg goes on to relate this to evolution.

by Greg Mayer

Although Jerry posted the following video a couple of years ago, it came to my attention again yesterday, when a friend sent it to me. (And I did not recall until I checked that Jerry had posted it!)

My friend asked, “Can your cat do this?”, to which I replied

“Yes, if you taped her sitting at the keyboard long enough, only selected those bits where she hit several keys in a row, and then had the orchestra play around these selected moments.”

What immediately came to my mind (and what I quickly tried to explain to my correspondent), was that the cat playing the piano was not the result of the cat “knowing” how to play the piano, but rather the result of a cumulative selection process, in which the cat’s more or less random key strokes and rubs are filtered for those that are “good”, and the good ones then strung together.  If you let the cat sit at the piano long enough, recording all the while, then splice together all the times it made several euphonious keystrokes in a row, you can build up a “solo”. The composer then composed a piece around these selected euphonious elements.

The video exhibits something akin to Richards Dawkins’ “me thinks it is like a weasel” story, which he related in The Blind Watchmaker (my favorite of his books, though I’ve not read them all). Given enough time, a monkey pounding at a typewriter would reproduce all of Shakespeare, but it would take a very long time indeed. But if you allow cumulative selection to work—saving correct steps when they occur—it is possible to get a coherent phrase rather quickly. Dawkins illustrated this with a famous line from Hamlet, in which Hamlet is making a fool of Polonius; says Hamlet, “Methinks it is like a weasel.”

The probability of a monkey producing the 28 characters in the sentence in a single try is one in 27 (the number of letters plus the possibility of a space) raised to the 28th power, or roughly 1/10^40– a mind-bogglingly small chance. But if you select any correct letters that happen to appear, and then let incorrect letters vary again, and then repeat, you will soon get the full sentence. In Dawkins’ first try with a simple computer program that implemented this selection algorithm, it took just 43 trials (“generations”) to get it, and that result was typical. The point of course, is that random variation and cumulative selection is a very different process from just random variation (which many critics of natural selection seem not to get).  (The program captures only some of the characteristics of cumulative selection, and Dawkins discusses these caveats in the book: see Chapter 3, “Accumulating small change”).

In the “Catcerto”, the keystrokes of the cat (which are apparently encouraged in some way by her owner, whose hands appear briefly at one point in the tape) are recorded, and the euphonious combinations selected, much as the correct letters are saved in Dawkins’ program. The composer can then select from among these, and splice them together, including changing their order (something for which there is no analogue in Dawkins’ program), and then write the chamber orchestral score around these spliced together euphonious moments.

h/t: D. Pham

20 thoughts on “Caturday felid: “Methinks it is like a catcerto”

      1. Well, one of Reagan’s speechwriters included the “take up arms against this sea of troubles” bit without realizing this phrase referred to suicide.

    1. Actually, I bet he’d approve heartily. This would be an awesome exercise for composition students: take a seemingly incoherent jumble of notes, and make it the centerpiece of a meaningful work of music. For bonus points, use the same “theme” in multiple styles, from Baroque though Serialism.

      Mindaugas Piečaitis, obviously, would receive the highest marks possible for such an exercise; the Catcerto is superlative.



  1. It’s sort of how I started to learn the violin. I would just scratch away at random at first then keep repeating the bits that sounded less bad. I got ‘twinkle twinkle’ in a few minutes.
    The Bach Partita in E major will take a little longer.

    1. If you are inclined to improvisation, a mix of random, arpeggios, and repitition can occupy a lifetime of musical enjoyment.

      1. a mix of random, arpeggios, and repitition can occupy a lifetime of musical enjoyment.

        If observed from a suitably large distance … (says me, who failed years of piano tuition, through not being in the slightest bit interested).

      2. If you play bass guitar learning the minor pentatonic scale and the six tone blues scale are pretty much all you need. Just play notes at random in those scales and you can come up with anything. I found the Hawaii Five-O theme in the minor pentatonic. The minor pentatonic and blues scales are transposable on bass, so you can play in any key with the same fingering. Major scales are a bit harder, but I have managed to compose a transposable tab for Happy Birthday when I was doing a cover version of The Call’s “When the Walls Came Down”. Yep, that entire song is a warped version of Happy Birthday.

  2. I’ve never liked the weasel program as an explanation or example of natural selection. An ID proponent will at once say “This is intelligent design! You had to tell the program in advance what result to aim for.”

    1. Which, of course, Dawkins discusses. The weasel program is much more like artificial selection, and effectively counters (as it was intended to) the “tornado in a junkyard” model of evolution favored by creationists since, well, since forever. The Catcerto is more like natural selection, because there is no “goal” toward which the selection process is moving. Rather, the variations arise randomly, are retained for their euphony (correspondence to the conditions of existence), and thus accumulate, so that we don’t know where the endpoint will be when the selection process begins.


      1. To avoid that criticism, one could presumably modify Dawkins’ example by using the same process to arrive at any sentence from Hamlet, rather than one in particular.

        Of course, “telling the program in advance what result to aim for” doesn’t require intelligence. The environment does provide “goals”, or at least conditions which greatly limit the scope of what traits can be successful.

        1. Right.

          Giving the program a constraint in the form of a goal sentence is analogous to nature giving finches and their beaks a constraint in the form of what food is available.

          I think the analogy is still apt. I suppose it could be slightly improved by removing the human element and letting a random number generator give the program a goal sequence of numbers.

      2. Or in a deterministic system pseudo randomly. If our choices are implicit in the initial conditions of a deterministic universe, then so are all evolved forms.

    2. This discussion reminds me of going through calculations years ago about the relative efficiency of different sorting methods in computing science. Quite small changes in the algorithm can have considerable changes in the effectiveness (speed, memory requirements) of one algorithm compared to the other.

  3. > But if you select any correct letters that happen to appear, and then let incorrect letters vary again, and then repeat, you will soon get the full sentence.

    This is an incorrect description of the program. Correct letters are not selected and fixed (that would be far too easy). Correct letters merely improve the score of the candidate, but can mutate again. Such undesirable mutations lower the score, however, and are therefore unlikely to be selected for the next generation.

  4. A few years ago I got into a coffee shop debate with an “animal whisperer” who argued for an extremely high intelligence level for cats. That is, she had claimed that she sat down with her cat, explained how much it bothered her when he brought dead mice into the house, and her cat had understood and stopped doing this. My attempts to offer alternative explanations and deal with what we really know about the cat — they do not think and understand language just like humans — fell flat.

    She later sent me a video of an elephant painting a flower, crowing that this was clear evidence that I was wrong. Of course, the paintings were the opposite of what has occurred with the catcerto: lots of operant conditioning and prompting.

    But no. My guess is that this particular “animal whisperer” (and everyone else in the room, who agreed with her) would see BOTH the cat and the elephant as exercising genuine creativity and human-like discipline. What you see with your own eyes just can’t be denied!!!

    Skeptics are intellectual snobs taking magic out of the world and hating on animals.

Leave a Reply