Wednesday, February 18, 2015

Much Ado about Simulation (Characters as Props)

When I first started posting at The Valve I posted a series on the problem of literary character: Since they ARE fictions, why is it so difficult for us to talk about them AS fictions? Why are we always using that language and concepts of real people to talk about these fictions? This is one of those posts, originally going on the web on August 5, 2006.
Let's look at the individual reader as he or she apprehends a text and thus (re)creates the lives of the fictional characters in the text. It is common to say that we come to identify with literary characters. But, as Norman Holland pointed out in The Dynamics of Literary Response (1968, pp. 262 ff.), it is by no means clear just what we mean by identification in this context. Still, in order to get this discourse on the road, we need some word for the relationship a reader establishes with a character. If not “identification,” then what?

Keith Oatley has been writing about fiction as simulation of the world:
Shakespeare’s great innovation was of theatre as a model of the world. The audience member constructs the simulated model in the course of the play, and thereby takes part in the design activity. So fiction is to understanding social interaction as computer simulation is to understanding, perception and reasoning. Shakespeare designed plays as simulations of human actions in relation to predicaments, so that the deep structure of selfhood and of the interaction of people who have distinct personalities becomes clearer.*
Oatley has the notion of simulation from computing, where computers are used to simulate a wide variety of phenomena – traffic patterns, explosions, fluid flow, and so forth. He proposes that simulation is just the notion we need in order properly to interpret the Greek mimesis. Stories are “the kind of simulation that runs on minds rather than on computers."

I find Oatley’s proposal to be plausible, but I’ve got reservations. Thus much of what I say will be a critique of that notion. I am not particularly happy about this mode of proceeding, as I would prefer simply to set forth a problem-free account. Alas, I am not aware of such an account and so must be content with a crude demonstration by via negativa.

A Scene from Shakespeare

I would like to discuss Shakespeare’s Much Ado About Nothing, focusing on the first scene of Act IV. I have two reasons for choosing this scene: 1) Eight major characters are on stage and most of them have substantial speaking parts in the scene. 2) The scene is emotionally rich, with the characters having distinctly different interests in the action. Whatever it means for a reader to simulate an imaginary world, the complexity of this scene taxes that capacity.

Here’s what’s going on: The characters have gathered for the wedding of Claudio and Hero, the arrangement of which has been accomplished in one of three plot lines intertwined in the first three acts. The relationship between Beatrice and Benedick is another of those plot lines. While the third is Don John’s scheme to destroy the wedding. Don John’s notion has been to deceieve Claudio into believing that Hero is a woman of loose morals who has deceived him even after having accepted his proposal. Thus, while the other characters, save Don John, believe they are about to witness a wedding, Claudio intends to denounce Hero before the assembled group.

And that’s what he proceeds to do, within thirty lines of the scene’s opening. Hero doesn’t say much, but she does deny the charges. She then faints and is taken for dead. At that point Don John, Claudio, and Don Pedro leave. Hero then revives and those who remain plan a course of action that will, they hope, clear her name.

My first issue is this: Is it physiologically possible for one person, one nervous system, to simulate the actions and emotions of all the characters in this scene? Different emotions are mediated by different neural and physiological systems. In particular the sympathetic and parasympathetic systems are important in motivation and emotion and they are antagonists, pulling physiological processes in opposite directions. Claudio’s aggressive anger – perhaps with overtones of hatred – would be sympathetically driven while Hero’s protective faint would be parasympathetically driven. Can a single nervous system simulate both of those states, either simultaneously or in close succession? That seems highly unlikely. And those are only two characters in the scene. What of Hero’s father, Claudio’s patron, of Beatrice and Benedick, the Friar? And what of Don John? Is he feeling pleasure, perhaps even triumph – albeit concealing these feelings from the others – at seeing his plotting bear fruit?

It seems rather unlikely that a reader would be able to simulate these various feelings and attitudes within the relatively brief compass of a few minutes. Beyond the difficulty of simultaneously activating mutually exclusive neuro-physiological systems, we have the fact that these hormonally rich systems change state more slowly than perceptual and cognitive systems. Even if we simplify the reader’s problem by asserting that the reader only need simulate the character who is speaking, we have the problem of switching from one character to the next, which could be daunting where three or four characters switch back and forth within the compass of only a dozen or two lines.

So, perhaps the reader does not simulate these feelings and attitudes in any very deep way; in particular, perhaps the slower acting hormonal systems are not recruited into action at all. Or perhaps the reader is not simulating the emotions of any of the characters in the scene. Rather, the reader is simply reacting to the actions and words of people whom the reader “knows” and toward which the reader has various attitudes, both positive and negative. That is, if the reader is simulating anyone, it is a person watching such a scene. I am imagining that the reader is simulating someone in attendance at such a wedding, but not participating in them in any way.

And this is not so far from imagining the reader to be in the audience of a performance of Much Ado About Nothing. In this situation each actor has responsibility for simulating the words and actions of a character, and only one character. The playgoer need only react to the play.

At this point, however, the notion that a reader, or a playgoer, is simulating the action seems rather far away from what Oatley has been asserting, who talks as though the reader is simulating the characters from the inside. Though he doesn’t use phrases like “from the inside,” that seems to be what he is asserting. If, for the reasons I have asserted, that is difficult or impossible, then it is not at all obvious to me what simulation might mean. What could it mean to simulate a character from the outside?** (Note that this problem doesn’t arise in computer simulation of physical phenomena.)

One way to deal with this problem would be to say something like: “Well, we don’t simulate all the characters. Just one or a small group of them.” Given that Oatley is arguing that such simulations help us understand ourselves and others, and thus help us negotiate our social lives, it is not at all clear that such a narrowing of scope is legitimate. But even if we accept it, difficulties remain.

The Foolish Protagonist

Let us return to Much Ado. Unlike Hamlet or Othello, the play doesn’t really have a single protagonist. But Claudio takes the active role in one plot and is obviously is a central figure in this drama.

Let us say that Claudio is motivated by anger in this scene. But the accusation motivating that anger is wrong, and the reader knows it. That knowledge effectively bars the reader from “identifying” with Claudio and so simulating his anger toward Hero. If the reader feels any anger at all in this scene – as this reader did – it is more likely to be directed at Claudio himself, perhaps against Don John as well, or simply at the whole bollixed situation. One sort of reader may also feel a bit of pity for Claudio, who, after all, was deceived; while another sort of reader may feel that Claudio was wrong not to have first broached the matter in private. But no reader is simply going to follow along with Claudio’s feelings and actions.

At this point, it seems to me that, if the notion of simulation is to be of much use, that we need to know considerably more about just how the brain does these things. Rather than speculate about what such knowledge might yield, I want to move in a different direction.

Some Hypotheticals

At this point, with the contrast between what the reader knows and the characters know, we have entered conceptual territory more familiar to discussions of the novel. So let’s wander about in that territory and imagine how this same story could be staged so that this particular episode would affect the reader in a different way.

Let’s imagine that Shakespeare had written the play in such a way that, during this scene, the reader knows no more about the situation than Claudio does. Thus, whether or not the reader thinks Claudio is right in taking this particular action – public denunciation – the reader believes that Claudio has a case and certainly must do something. I can imagine two ways of doing this.

One strategy would be to begin the play as Shakespeare does, but hide all of Don John’s plotting. One might, as well, actually show the scene where Claudio sees Hero with another man – Shakespeare doesn’t do this, but only has Conrade and Borachio talk about it. This would have to be done in such a way that the reader (or theater-goer) is as deceived as Claudio. In this version the reader would be more readily able to “identify” with Claudio’s actions in the wedding scene. Of course, the reader would also be privy to the plotting that transpires after Claudio leaves the scene along with Don John and Don Pedro. The truth about Hero would have to be revealed in subsequent scenes, which would require some arranging, but would certainly be possible.

Another strategy would be to open with the wedding scene. In this case the reader would have no prior knowledge of these characters and thus no prior commitment to any of them. With respect to the reader’s belief in what happens in the scene, this is not very different from my first proposal – though rather more back-story will have to be revealed in subsequent scenes (cf. the way Sophocles staged the Oedipus story). In neither case does the reader know that Claudio has been cruelly deceived. But in this version the reader knows nothing about anything, but, experienced reader that she is, she also knows that the whole story lies ahead and so might be less inclined to take events at face value.

In neither case, however, does the reader absolutely know Claudio to be wrong. It thus seems to me that the reader’s experience of the wedding scene in either of these hypothetical versions will be very different from the reader’s experience of the scene in the play as Shakespeare wrote it. Yet the characters – their feelings, attitudes, desires, and actions – are, by definition, the same in each of the three versions. To the extent that we, as critics, analyze and conceptualize the action in a play as unfolding from the desires and motivations of the characters, the difference between these three different ways of telling the story is irrelevant. It has no bearing on what the characters are doing. That it does affect how readers experience the story is beside the point. The story is the story and that’s that.

Can we be satisfied with such a position? I think not. Then, what’s the point in telling the story one way, as opposed to another? That seems to me to be a subtle matter, one I’m not prepared to address at the moment. So I must rest content with the assertions that the difference does matter and that we’re not going to figure out how it matters by analyzing the characters. We must look to the interaction between the text and the reader.

Simulation, Again

Let’s return to the notion of simulation. Rather than thinking about reading as simulation, however, I want to think about how one might simulate a set of actions, such as those in Much Ado About Nothing, using a computer. I have no idea when, if ever, we will be able to do such a thing. But for the moment let us assume that we can do it.

The obvious thing to do is to treat each character in the play as an autonomous agent and to then develop a model for each character. When that is done one can then initialize each of these models to a state appropriate for the story in Much Ado About Nothing and run the simulation. The running of the simulation would consist of having the character simulations interact with each other until an end state is achieved. In such a simulation one is not going to write code that says that Claudio will fall for Hero when he sees her, Don John is going to plot a deception when he finds out, and so on. One writes models of characters and then has them interact on their own. No more, no less.

When we set the simulation in motion, however, we do not know how the characters will in fact interact. We’ve done our best to create the models so that, given the appropriate initial state, the simulation will take a certain course. But that course is not guaranteed. That’s the point of running computer simulations. We don’t know how a system will evolve given some initial state and so we create a simulation so we can explore the system’s state space.

Thus when we first run our simulation it might enact a different story. This, in fact, seems the most likely thing. Perhaps the story is different in a minor way – Don John has Fred the Grocer play the paramour instead of Borachio. Or perhaps the difference is major: Don John’s plot is exposed before the wedding, which goes off without a hitch; but Don John elopes with Beatrice and Benedick hits on the Friar. If the difference is major, then we have to tinker with our models and with the initial conditions until we get a run that enacts the correct story. That might take quite a bit of tinkering. Presumably all that tinkering will teach us something.

This sort of simulation is quite different from what Oatley proposes. Of course, I’m talking about running a simulation on computers while he is talking about fictions as simulations running on brains. But, from my point of view, that is a minor difference. Why do I say it is minor?

Well, if we have the computational capability for running the kind of simulation I propose, then we would also be able to simulate someone reading Shakespeare’s text. In this case, the characters in the play would not be autonomous agents interacting with one another. They would simply be simulations of whatever it is that happens when real people read real texts about imaginary characters. That's all I want to know. No more, but perhaps I'll be satisfied with only a crude approximation.

Sp What?

Indeed. If I had a nice conclusion ready and waiting I would be writing a formal paper instead of putting some notes in a bottle and tossing the bottle into the blogosphere. I’m inclined to think that the way to proceed from here, given this particular example, would be to ponder the difference between the story as Shakespeare has staged it and the hypothetical versions I’ve suggested. Shakespeare’s strategy forces “distance” between the reader and a protagonist at a critical point in the story. Why?

Given an answer to that question, what then? This story is only one story, and Shakespeare’s play is only one version of it – as you know, almost all of his plays are but one version of a story that had been told in other versions. How do we get from this one case to the more general question of fictional characters?

*Shakespeare's invention of theatre as simulation that runs on minds, URL:

**See the concept of vicarious experience in William Flesch, Comuppance: Costly Signaling, Altruistic Punishment, and other Biological Components of Fiction, Harvard University Press, Cambridge, Massachusetts and London: 2007. 252 pp.


  1. It seems to be a major requirement of Shakespeare's poetics that the reader be given topsight over any character—that no secrets are kept from the reader, or rather that the reader be not deluded along with one of the major characters. One might surely find counterexamples, but they are minor ones.

  2. Shakespeare's baroque poetics invites the reader or theater goer to both know the truths and see the fatal existential effects of false beliefs – so the emotional impact has to do with being torn between the two positions of the truth-knower and the character that lacks access to relevant truths. One could call it an essentially epistemic dramatic game: seeing people suffer from not-knowing calls for empathy, since everyone has experienced this in life. (Had I but known…)

  3. As I'm sure you know, Harold Bloom has made extravagent claims on Shakespeare's behalf. I've not read his book, though I've glanced through and read a somewhat earlier and much shorter version of those claims, and I'm sympathetic to the idea that Shakespeare, in some sense, created us. And how he did is right there in his poetics, in getting us to sympathize with characters who are subject to deep and perhaps self-generated misunderstanding and thereby cause considerable pain to others and ultimately themselves.