In this post and another I am going to gather a few thoughts about what it has been like trying to do experiments in collaboration with colleagues from psychology faculties. This has been a brilliant experience, but it hasn’t led to a concrete clinching result, now published in Nature. I am going to be vague about some of the details, and will be writing about a sort of merged amalgam of multiple attempts. The next post, which will be entitled ‘Some Things I Learned From My Experiments (2)’, will actually be partly about why I am being vague in this way, rather than telling all.
This one is about the consequences of trying obstinately to focus on the things I would normally want to focus on. To explain what I mean by that, first I want to cite something that, to good effect, did not take the same approach. As far as I can tell, I have never mentioned a recent essay that definitely counts as a proper collaboration between literary critics and experimental psychology: it’s ‘Cognition, Endorphins, and the Literary Response to Tragedy’, by Felix Budelmann, Robin Dunbar, Sophie Duncan, Evert van Emde Boas, Laurie Maguire, Ben Teasdale, and Jacqueline Thompson (Cambridge Quarterly, 46 (2017), 229–250). This is a major omission, because it’s an important piece of work.
They gathered a large audience (which I now know is a great achievement), got them to watch a film that all concerned agreed was a tragic story, and then measured their performance in a test of physical endurance. With appropriate controls and comparisons in place, they found that watching tragedy did actually increase people’s ability to put up with discomfort. So they offer an endorsement for the long-theorized idea that watching suffering affects our emotional make-up in some positive way. I like so much about this: it’s resourceful, thought-provoking work. However, it’s not really what I want to do myself. I have always wanted to focus on details (of the text, or of the play in performance, etc.) in my attempts to construct experiments, and this has — so far — led to intriguing processes but murky outcomes.
I once spent a long time trying to show that a particular feature of the language of a Shakespeare play had a psychological effect on its readers and audiences. I would always, in my critical writing, be investigating at that level, rather than describing the effects of a play as a whole, and I, aided and abetted and inspired by others, wanted to carry on in that vein. We decided to see whether a film of the original version of a scene, and a film of the same scene ingeniously doctored to remove that particular feature of language, produced different results in an audience questionnaire. Although it is not the point of this post, I will tell you that it did somewhat, but not in a statistically significant way, and not encouragingly enough for us to redo the whole thing on a more massive scale.
One reason why we stopped at that point was an awareness that the thing we had produced, the doctored text that removed the feature of the text about which we were hypothesizing, was a bit of a monster. On the face of it, it was cleverly done, absolutely recognisable, fitting into the play perfectly, but we never entirely believed, or looked forward to having to persuade anyone, that removing the whole of, say, one kind of trope, or one formal feature, or one semantic field, was really the equivalent of an experimental variable. The language of a literary work isn’t composed of discrete types of thing, such that you can remove one without complex consequences, not really traceable, for the rest. This paradigm — testing the effect of a textual feature by comparing the original to an altered version that eliminates the feature — was always going to involve some compromise, and a risk of misrepresenting the nature of a literary text.
I suppose a more waspish person than I might say that this is sort of true of a great many things treated as discrete variables in psychology experiments — the unintended consequences of the interlocking nature of everything are a bit of a haunting presence. However, there are conventions that allow variables to be isolated in cognitive science, and there really aren’t in literary criticism, and I don’t think they’re particularly desirable anyway. So, yes, the fact is, I really wanted to get somewhere by isolating a feature of a text as a basis for an experiment, and it was an interesting and entertaining process, but didn’t turn out satisfactorily.
I should note that the thing we focused on, and removed, was designed to speak to an idea being explored by a psychologist collaborator. Their take on it was optimistic, and they saw what we were trying to do. However, the question of the truly discrete variable was a problem from the psychological side too, though this was more focused on the overall experience of the films we made: were they really identical apart from the textual variation? Not absolutely, of course, and how satisfactorily? This post isn’t meant to sound melancholic. Hope springs eternal. But literary nuances as experimental variables: they’re a puzzler.