Creativity, Part 2
"The greatest scientists are artists..." (Einstein)
"All of the great entrepreneurs are artists" (Naval Ravikant)
"Michael Jordan was an artist..." (B.J. Armstrong)
Success in most fields requires more than just knowledge and skill. Artistry is needed. Great scientists, entrepreneurs, and athletes work outside the box, generating innovative solutions to each obstacle they encounter, whether that obstacle is a mathematical paradox or Shaquille O'Neal.
In a word, they're creative.
"Creativity" in this sense is a relatively new term, but people have long recognized the value of original, useful ideas. And, we've long complained that our natural creativity is stifled by school, work, and social convention. Rousseau, for instance, had much to say about that in Emile two-and-a-half centuries ago. The modern version of the complaint is typified by Ken Robinson's popular Ted Talk "Do schools kill creativity?". "We're now running national education systems," he says, "where mistakes are the worst thing you can make. And the result is that we are educating people out of their creative capacities."
People like to say that creativity is essential to human progress. Ironically, we're also calling for creative solutions to crises that ultimately stem from progress (climate change, pollution, overpopulation, diseases of affluence, AI disruptions, etc.) No surprise, then, that we're constantly seeking ways to boost creativity. The internet is buzzing with resources, paid and free, that claim to help. I'll say more about them at the end. My focus will be on a new study, published two weeks ago, looking at whether marijuana in particular makes people more creative. I love this study – the design is exceptionally creative – but it also illustrates how statistics negatively impacts how we understand ourselves.
Marijuana and creativity
Some people believe that marijuana makes them more creative. Is that true, or are they just high?
Entire books have explored connections between drug use and creativity; critics have pushed back. It's clear, at least, that if drugs do boost creativity, the way that it happens would depend on the type of drug.
Psychedelics like LSD and psilocybin are said to enhance creativity mainly through what psychologists would call altered cognition. When you're under the influence, you experience heightened awareness of sensory details, distortions of perception, unexpected associations, unusual perspectives, etc. – crudely speaking, you're thinking outside the box, even if you're not trying to.
In contrast, the benefits of marijuana are typically linked to changes in mood rather than cognition. Some of the most successful, creative people, including Maya Angelou, Louis Armstrong, Lady Gaga, Steve Jobs, Stephen King, Bob Marley, Rihanna, and Carl Sagan, have acknowledged marijuana's role in their successes. But they typically spoke of enhanced mood rather than altered cognition. Under the influence, physical and mental discomforts were eased, anxiety dissipated, and they were able to focus their creative energies, perform better, and so on.
Rationale for new study
The new study, published two weeks ago in the Journal of Applied Psychology, also assumes that if marijuana boosts creativity, it does so by affecting mood.
Specifically, Dr. Yu Tse Heng and colleagues at the University of Virginia, University of Washington, and National University of Singapore assumed that low levels of marijuana consumption make a person more open, cheerful, and sociable. The word they used for this is "jovial". In a jovial state of mind, you feel safe, playful, receptive, willing to take risks (creatively speaking), and you can then be more creative.
(One might ask then whether anything that makes you jovial will enhance your creativity, or whether there's something special about marijuana. The researchers don't address that, because, as you'll see shortly, it's a moot point.)
Study methods
As experimental studies go, this one was exceptionally creative. People who report light use of marijuana were recruited. The 191 participants were asked to complete an alternative uses task online (specifically, they were given 4 minutes to list as many creative, possible uses for a brick as they could think of – see last week's newsletter for more on this task). Judges were later recruited to rate the creativity of each response. Agreement among judges – i.e., interrater reliability – was quite high.
Participants also used 5-point scales to rate their own creativity on the task, as well as their happiness at the time.
Each participant was assigned to either a cannabis group or a control group. The cannabis group completed the task 15 minutes after using marijuana. The control group completed the task at least 12 hours after last marijuana use.
What makes this study especially creative is that it didn't take place in a lab. Participants weren't even told how much marijuana to use or when to use it. They were at home, and they chose when to go online and complete the task. They were simply asked to do so while sober, or immediately after getting high.
The strength of this approach is its naturalness. Mood and creativity weren't inhibited by the artificiality of a lab setting. Naturalness, of course, may also be a limitation of the study. We don't know how much cannabis each participant consumed. We don't know whether some participants completed the task under distracting conditions. We don't know whether they'd been consuming alcohol or other drugs as well. (I'm not saying that any of these things occurred. I'm just saying that we can't rule them out.)
In experimental research we usually see trade-offs between naturalness and rigor. The more rigorous the experimental set-up, the less natural the conditions. Here the researchers prioritized naturalness, which is arguably the best choice given the importance of mood as a variable.
Study findings
The researchers found that marijuana use increased joviality but did not affect creativity. In other words, the cannabis group felt more jovial than the control group did, but the two groups didn't differ in mean creativity ratings. However, the cannabis group rated their own creativity higher than the control group did.
The researchers conducted a second study using a more business-oriented task (coming up with revenue-generating ideas for a local band). The results were the same as before, with the additional finding that participants who'd used marijuana also rated other peoples' creativity more highly.
In sum, what we have here is faintly reminiscent of a Cheech and Chong movie. When people use marijuana, they feel happier, and they think they've become more creative. In fact they haven't. They're just happily impressed with their own creations – and everyone else's too. (Dude, this is awesome!)
Practical implications
This is a timely study, given that recreational use of marijuana is now legal in 22 states, as well as Washington DC and Guam, and legalization in other states seems likely soon. Two implications of the data seem important.
1. Marijuana neither improved nor impaired creativity. However, the study focused on low levels of consumption. Other studies show that higher doses of marijuana do impair creative performance. So, if you're interested in using cannabis to boost your creativity, proceed with caution.
2. The results were presented in aggregate. We know that the two groups (cannabis vs. control) showed no means differences in creativity, but means are ambiguous with respect to individuals. Any particular person in the cannabis group may have been positively or negatively affected. This suggests, again, a need for caution in using marijuana to boost your creativity. It might help, it might not.
Here we also see one of the most fundamental ways our society has been "statisfied", or transformed by statistics.
The limitations of the mean
The researchers' approach to data analysis is mainstream in its focus on aggregate data (i.e., arithmetic means). Nowadays, we rely heavily on means to describe people. However, prior to the 19th century, means were only occasionally used in scientific research. For instance, when astronomers obtained inconsistent measurements of the same phenomenon, they'd calculate the average and treat it as the true value. Otherwise, it was rare to find means in science (or anywhere else). Elsewhere I touch on how means came to be so prevalent in scholarship and in public discourse.
Means are informative, but they may conceal important individual differences. For instance, in this study, the cannabis group and the control group each had mean creativity scores of just over 8 on a 25 point scale. That's informative. It tells us that on average the two groups were essentially the same in creativity. But let's think for a moment about what we don't know. A mean of around 8 is consistent with each of the following scenarios:
Scenario A: Everyone in the control group, and in the cannabis group, scored an 8, or something close to that.
Scenario B: Everyone in the control group scored an 8, or something close to that. However, some people in the cannabis group obtained very low scores, some people scored around 8, and some people obtained very high scores.
These are vastly different scenarios. Scenario A tells you that marijuana simply doesn't affect creativity. (This is how the researchers present their results.) Scenario B tells you that marijuana does affect creativity in many cases, but that the effects depend on the person. Some people become a lot more creative, others become a lot less creative.
There's nothing special about this study. Lots of studies rely on aggregate data. But in many cases, as with this study, you want to know whether Scenario A or Scenario B is a better approximation of what's happening behind the scenes.
I reached out to Dr. Yu Tse Heng, lead author of the study, about this and other issues. Her response was quite revealing:
[Springer]: I'm curious about individual differences in your creativity rating data. The means for the cannabis and control conditions are pretty similar, but, as you know, means are consistent with various sorts of distributions. Would you say, for example, that with respect to creativity, almost nobody was affected much by condition, or were there signs that in one or both of the conditions, there were subgroups of people who had relatively high (or low) creativity ratings?
[Heng]: This is also a good question. While we randomly assigned participants to either the cannabis or the control condition, it could certainly be the case that individual differences might also influence creativity ratings. We did not pursue these questions in this research but agree that these are important questions worthy of further investigation. For example, it would be interesting to understand if patterns of cannabis consumption (e.g., strains, doses, consumption method) would affect one’s creativity. These parameters around cannabis use are currently challenging to test in studies due to federal regulations but could be a possibility if federal regulations ease up in the coming years. Overall, I’d say that our study is only a first step in understanding the relationship between cannabis and creativity, and much more research is needed to fully understand the complex and likely nuanced effects that cannabis use has on our various creative endeavors.
I'm grateful to Dr. Heng for her response. What's noteworthy about it is that when I mentioned individual differences, she immediately thought of differences in types of cannabis and consumption practices. I agree this is important. But I was asking about a different issue, one that could be addressed by analyses of the creativity ratings her team already has.
I'm not saying that Dr. Heng and colleagues did anything wrong. Their study is cleverly designed and relies on a mainstream approach to data analysis. My quarrel is with the mainstream. When you report means, you need to also distinguish between outcomes like Scenario A and Scenario B. (Standard deviations and other variability statistics don't necessarily help much.) Here's another possibility:
Scenario C: Everyone in the control group scored an 8, or something close to that. Half of people in the cannabis group scored a 1 or 2, while the other half scored around 16.
In other words, a mean of 8 is consistent with the possibility that half of people who take marijuana become much more creative, while the other half become quite impaired. (I don't have any particular reason to predict this. I'm just saying that we don't know what we don't know. Other scenarios besides A, B, and C are possible.)
Bottom line? it's a great study. It confirms that marijuana use increases joviality (on average). And, it tells us that, on average, light marijuana use enhances perceptions of creativity but not actual creative performance. But it provides very little guidance for individuals.
Can you increase what you can't measure well?
"Creativity" means different things to different people. This is evident from a quick search of both popular and scholarly advice on how to enhance it. Some advice focuses on fluency (generating a lot of ideas in a hurry). Most emphasizes divergence (coming up with relatively distinct, innovative ideas). Some experts, like Robert Sternberg, stress the importance of developing a creative habit of mind. Some advice is for educators, some is for entrepreneurs, some is for personal growth. None of it is likely to help much.
Last week I noted that psychologists admit to not having learned much yet about how to enhance big-c creativity (the kind that yields important creative products), owing in part to their use of measures like the alternative use task ("how many things can you do with a brick?"). Once again here's Dr. Heng:
While the alternative uses task is widely used in psychology research, it definitely has its limitations—one of which is, as you’ve pointed out, its ecological validity [generalizability to all sorts of settings where creativity could be used]. The task we used in our second study where we asked participants to generate creative ideas to help increase revenue for a local band is a more ecologically valid work-related creativity task. But again, there are limitations to how creativity research can be easily performed in the lab/experiments. For example, we would not be able to capture the creativity needed for scientific innovation and breakthroughs or for the creation of works of art from these two creativity tasks. We thus acknowledge the need for more research on this topic and encourage more work in this space.
That's nicely put. We know something about"little-c" creativity, such as the ability to brainstorm solutions to a routine challenge, but not much about the "big-c" creativity responsible for major innovations and creative products.
Isn't the measurement and enhancement of little-c important? Well, yes, but it has become increasingly clear that creativity in any meaningful sense is a social process and culturally-specific. Consider the work of Dr. Danielle Tyree, my favorite doctoral advisee, who graduated in 2020 and has gone on to a successful career in educational consulting. Danielle's dissertation focused on cultural differences in creative performance, and one thing her work tells us is that asking someone to list novel uses for a brick might underestimate their creative talents, if they're not used to questions like that, or if you're not giving them enough context. Danielle's work suggests that a better version of the task would be to embed it in a natural setting. For example: Suppose you're stranded on a small desert island with such-and-such geographical features. A brick washes up on shore. List all the novel, useful things you could do with that brick.
Arguably, this task still won't tap into what it takes to change the world via scientific work, entrepreneurship, etc. Psychologists have tried without success to predict big-c creativity, owing to the tasks they use as well as the assumption that divergent thinking is the most prized sort of response on a creativity task. As Danielle puts it "Straying from the norm is not celebrated in every social context like it is the American context." So, we may need more creative approaches to evaluating creativity if we wish to understand it better. This leads me to my final topic.
AI and creativity redux
Maya Angelou once compared talent to electricity, observing that "we don't understand electricity. We use it."
The same holds for creativity. This immediately raises the question of whether artificial intelligence could ever surpass us in creative performance. After all, AI doesn't "understand" creativity either, though, like us, it can readily generate novel ideas. Could AI programs ever outperform us in this respect, just as they're now better at mining large databases or writing certain kinds of code?
At least two studies, currently in press, have looked at AI performance on alternative use tasks. The results are mixed.
In one study, ChatGPT generated more useful but less original ideas than people did, and on the whole human creativity was found to be slightly superior. (Hooray for our team!)
In the other study, which compared people to five different AI chatbots (including ChatGPT-3 and -4), AI had the edge. Although the originality of AI and human responses were comparable – and mostly indistinguishable – the AI chatbots exhibited greater fluency (i.e., a larger number of ideas).
Is either study a cause for concern? I would argue no. Alternative use tasks are pretty narrow. You get credit for saying things like a brick could be used as a weapon, a door stop, a hot plate, etc. Indeed, these are the kinds of ideas ChatGPT comes up with. They may be novel ideas, but they're not going to change the world, and so it doesn't seem troubling that a program came up with them.
AI can support our creative endeavors, but it can't supplant that thing in us that, since the 1950's, we've referred to as our "creativity". This thing has gone by many other names, ranging from "the imaginative faculty" to "the voice of the muse". We've never understood it, but we're good at using it – much better than AI will ever be. Think of your favorite book, or piece of music, or scientific theory, or whatever, and I suspect you'll find some reassurance there.
Thanks for reading!