Is making a sentence like riding a bicycle? A neat little experiment by Victor Ferreira and his co-authors suggests, as I read it, that we store a verbal style and perhaps our very character the same way we store muscular skills like riding a bicycle.

Memory comes in several flavors. In semantic or declarative memory we remember facts about the world like ‘North Dakota has a larger population than South Dakota.' Procedural memory stores learned perceptual skills like reading a page and learned motor skills like swimming. The two kinds of memory differ: procedural memories are learned slowly and decay slowly, the opposite of semantic memory.

Now you would think--I would think, anyway--that syntax, our knowledge of how to put words together to form sentences, would be third-person knowledge, involving semantic memory like knowing the population of South Dakota.  Not so, says Ferreira's experiment.

To tease apart the roles of these two different kinds of memory in forming sentences, Ferreira and his co-authors compared the way amnesics who had impaired semantic memory made sentences with the way normal controls did. The experimental session had two key elements. First, the subjects repeated a sentence to prime a certain syntactic construction: "The governess made the princess a pot of tea" (double object dative). Then, after varous maskings, the subjects were to describe a picture. Would they be more likely to say "The boy is handing the guitar to the man" (prepositional dative) or "The boy is handing the man the guitar" (double object dative)? Then, they were asked if they had seen a given picture before or spoken a probe sentence before--this to test declarative memory. (And there were all kinds of other controls and maskings within each sentence-picture-sentence block.)

Both the normals and the amnesics tended (significantly) to use the sentence structure that they had been primed for. But the amnesics did significantly poorer in remembering whether or not they had repeated a given sentence before, because of their impaired declarative memory. Conclusion? Both normals and amnesics were storing sentence structure as procedural memory.

Now, that, it seems to me, is downright surprising. Syntax involves abstract, relational knowledge. How could that be stored the same way as, say, the skill of swimming?

But, if you think about it, doesn't swimming involve a lot of abstract relationships among muscles that it would take a physiologist to describe? Also mustn't all procedural skills share some abstract properties? This-before-that, this-during-that, if-this-then-that-but-if-the-other-not-that, and so on. You don't continue to pedal the bike when you put the brake on. You don't modify a verb with an adjective: she walked purple. Professor Ferreira writes me:

. . . if I learn some complex skill like riding a bike or playing the piano, is the memory system that underlies that skill a bunch of muscle memories all superimposed upon one another? Or do we extract something from all of the experiences so that we--or more precisely, our procedural memory systems--extract the structure of the domain underlying all of those experiences? Our data, by showing that something that's demonstrably abstract (syntax) can be stored by procedural memory suggests that at the very least, procedural memory has the capability of holding onto that abstract structure, rather than more superficial forms of memory.

The largest conclusion, then, that Ferreira et al. reach for is: procedural memory can retain "fully abstrct knowledge." Fine. But I'd like to push their findings farther. They are careful scientists, and I am not.

Their subjects were primed to follow a syntactic pattern that they had experienced before. But outside of the lab, we all of us have our preferred syntactic patterns. They form part of our personal verbal style. On the basis of this experiment--and I know it is only one experiment--I'd suggest that I retain my personal verbal style as a procedural memory just like my personal style of walking, swimming, or making love.

I would go even farther. Ferreira et al.'s experiment tends to confirm earlier work by Grigsby and Hartlaub (1994) and Grigsby and Stevens (2000). They suggested that our defense mechanisms and our patterns of relationship with others rest on procedural memory. In short, these experiments taken together suggest what we think of as character or identity (the terms I prefer) or personality (the usual psychological term) reside in our brains as procedural memories.

We've come a long way: from pouring the princess a pot of tea (double object dative) to my describing human nature. to you (prepositional dative).  But, hey, why not?

Psychological items I've referred to:

Ferreira, Victor S., Kathryn Bock, Michael P. Wilson, and Neal J. Cohen. "Memory for Syntax Despite Amnesia." Psychological Science 19.9 (Oct. 9 2008): 940-46.
Grigsby, Jim and G. Hartlaub. "Procedural Learning and the Development and Stability of Character." Perceptual and Motor Skills 79 (1994): 355-70.
Grigsby, Jim and David Stevens. Neurodynamics of Personality. New York: Guilford Press, 2000.

Most Recent Posts from This Is Your Brain on Culture

How Novels Make Me a Murderer

I want the killer, thief, terrorist, to succeed--why?

Who Says Film Is Dead?

Are movies dead? Not if you can help it.

All about Brain Scans

Find out all about brain scans—or almost all.