Late last millennium, in the limestone clad psychology building on 10th street in Bloomington, I tried to prime. This was a project growing out of the concerns in the Niedenthal lab at the time. She, and all her minions had spent a great deal of research time looking at how emotional state influenced processing of emotional materials, such as words and faces and categories. The participants were made emotional from listening to music, or watching movies – it always worked. Then, they did some kind of cognitive task – like detecting words, or categorizing expressive faces on gender, or deciding which one of two concepts fit better with a target concept, and so on. What we tested were, in part, emotion congruence – you are emotional, items with the same emotional meaning are detected and processed faster. We also tested what Paula called emotional response categories: You are emotional, you group things according to emotional response. For example, if we had made you happy by having you watch Benny and Joon and Hoosiers, and then listen to Mozart Eine Kleine nachtmusik, you were more likely to say that “puppy” and “parade” went together than “puppy” and “beetle”. You were also more inclined to place “cancer” and “divorce” together than “cancer” and – oh, some kind of other tissue like thing. The emotional went together. You did similar when we made you sad with Terms of Endearment and Sophies Choice, and when we scared you with Silence of the lambs.
But, we always got this critique – well, Paula said we always got this critique – it wasn’t the emotion, the feeling, that did the thing, it was the emotion concepts that we had primed by making our participants emotional.
We didn’t believe it. But, it needed testing.
The first attempt is published in the 1999 Psychological Review paper, Emotional Response Categories (Niedenthal, Halberstadt and Innes-Ker). The strategy we used there was kind of a negative strategy. Expose the participants to the emotion inducing materials, but remove the emotion. Hence, we had them watch the same emotion-inducing movies, but asked them to keep track of the point-of-view shifts (this was prior to Gorillas hiding in plain view). This neutralized the emotion – at least on our standard manipulation check. They behaved differently on our triads task – in fact all of them used emotion as the response category more often now, including those in the neutral condition. Hmmmm. We checked if they understood the movies and the feelings the people in them were feeling, and they did. Whatever they did, it didn’t look like priming, though, but, well, it still could be.
So, Paula instigated this research, that I did a great deal of the ground-work on. After all, I was the new grad student, and needed to get my chops up. (My first five studies…. Well…Let me just say, at least the emotion induction worked.) We were going to try to prime the emotion concepts, and compare how people did when primed to how people did when emotional on that same triads task. And, how were we to prime? Use a sentence-unscrambling task. That seemed to be tried and true. Gender stereotypes could be primed that way, for example. So, I set up to create sentences that could prime the emotion concepts of happiness and sadness.
First, what would we think those concepts consisted of? I consulted appraisal theorists. I consulted a massive survey by Scherer where they looked for prototypical emotional events. My sentences were constructed so that they described prototypical happy or sad events, conforming to the appraisal dimensions of that event, involving words that belonged to sadness or happiness, but that would not appear on the later test – the triads task. Five words for each sentence, where a sentence could be made out of four of them, with one left over. For the sad, the left over was also sad. For the happy it was happy. But, priming also means you can’t hit them over the head, so the task had to have neutral fillers.
And, once I had my three conditions of sentence unscrambling, I set out to pilot test two things: that they primed the emotion concepts, and that they did not make people emotional. The second was pretty easy – they don’t. (Of course, that is kind of a null, but you can compare that to what happens when we do the movie/music induction). The first one ended up being…elusive.
I started out, if I recall, with creating a few stories that could end either happily or sadly, and the participants had to circle the one that they found the better one. The ending was one word, so a choice between two. (If I recall right. I should have all this in one of my folders, if I moved it over from all those CD’s I burned when I moved country). Originally I wanted to get some materials from another researcher, and I wrote him, only to get a reply back that he did not have time to dig them out.
Really, when someone asks me these days, I’ll happily dig out and mail, if I can just find them. I actually mailed the sentences to someone about half a year ago.
I think it looked promising, but not overwhelming, and then, kind of did not look like something at all. Anyway, that is not the one that we discuss in our 2002 paper. Paula also discovered another measure, where you could create matrices of words, where you could embed real words to be found – you probably have seen them in the Sunday papers or cross-word journals, or kids puzzle books. The idea was that if you were primed with happiness, you would find more happy words, and with sadness more sad words. I think we tried this twice. No dice.
We altered the proportion of happy and sad sentences in the two conditions – in fact to the point that at least some noticed that the sentences in the sad version seemed to describe awful stuff.
In the end, we went with a lexical decision task, where we find weak evidence of priming. Emotion concepts can be primed. Kinda. Barely. Weakly. We think. Or possibly even not, now in hindsight, and 11 years after the publication.
And, when we compared people in priming vs. emotion condition, people behaved quite differently on the triads task. They also behaved differently on an ambiguous situation task. Emotional state pushed the participants around (well, their subsequent cognitions), priming did not.
During the time, we were wondering how on earth they got those neat priming results, because it sure wasn’t easy, and we ran 30 people in each condition (90 total). Perhaps the Hoosier students are different. Or I wasn’t very good at doing this.
Once in Lund, I became a rescue advisor of a student who had attempted to prime emotions for her masters thesis. She wasn’t looking to test if it could, her original advisor had just assumed it could be done. Evidently he hadn’t read our paper (well, not that I expected many to do, but it had been published prior to this thesis). And, yes, she did not get any effects either (though I’m not sure what kind of task she used).
And, today, I tweeted with Rolf Zwaan who is wondering about the large effects people seem to get with this task, so I just had to let him know that not EVERYBODY does get that at all, and linked our paper.
I just thought I’d go overly honest methods, and describe a little bit more about what happened that ended up on the cutting floor.