The Incredible Afterlife of Patrick Gunkel's List of 638 Personality Traits

One of the first experiments Patrick Gunkel conducted for his Ideonomy Project in the early 1980s required the development of a massive list of 638 personality traits.

A description of the “Human Kaleidoscope” experiment is too long to provide here, but you can read Gunkel’s elaborate and esoteric description on the MIT website. The 638 traits are divided into positive, neutral, and negative categories and include both very familiar traits (Confident, Witty, Anxious) and uncommon or unusually phrased traits (Insouciant, Nihilistic, Unfathomable).

Suffice it to say, the list of personality traits was a lengthy “organon” or list that Gunkel used to demonstrate the fundamental principle of idea chemistry or “idea combinatorics” that is the central empirical methodology of his science of ideas.

Personally, I believe this experiment needs to be re-conducted, re-framed and re-written in terms that academics can accept, and published in a peer-reviewed journal.

Emails I found in the MIT Distinctive Collections archive show that Gunkel’s friends tried to convince him of exactly this, in fact, in the early 1990s, but Gunkel refused to play ball. Most of the work has therefore languished in obscurity on MIT’s website ever since.

Yet not every aspect of Gunkel’s Human Kaleidoscope experiment has remained unknown.

In fact, Gunkel’s list of 638 personality traits has received considerable attention from academic researchers over the last decade—including researchers in artificial intelligence, including researchers from Facebook’s AI Group.

Facebook and other researchers have cited Gunkel’s list as a valid source of personality traits that can be used to do things like train AI models on human emotions or anticipate weaknesses in cyber security planning.

At no point have these researchers demonstrated any knowledge about the original intention for the list, ideonomy, or Gunkel himself.

I’m fairly stunned by the implications here.

Gunkel’s list was seemingly subjective, i.e., the result of brainstorming with a thesaurus, and yet it was comprehensive enough for others to use for their research.

Researchers probably noticed the list was associated with MIT, assumed “the Ideonomy Project” represented some kind of comprehensive peer-reviewed study, and went off to the races to use the list in their own research.

In doing this, i.e., in establishing an academic paper trail between Gunkel’s list and legitimate research, I believe the academic community has in fact demonstrated one of Gunkel’s key points about ideonomy—the science of ideas he invented—which is namely this:

Yes, that’s right.

That’s the point.

Lists are a form of qualitative empirical evidence.

It sounds almost absurdly simple.

And it is simple.

But in practice, in execution, the point is both profound and difficult to achieve.

Despite its monolithic appearance as an arbiter of fact and truth, agreement in the scientific community is often hard to find when it comes to the details.

By focusing effort on creating comprehensive canonical lists and/or taxonomies that define the scope of studies within a field, Gunkel was suggesting that it would become much easier to clarify the aims and objectives of scientific disciplines as well as the areas of disagreement. Lists provide a form of ground truth that is, in many cases, somewhat lacking, especially at the bleeding edge of research.

List-making and taxonomization would help advance research by providing an empirical jumping-off point for inter-disciplinary, cross-disciplinary, or multi-disciplinary work.

One of the most difficult things to do as an academic researcher is to start a conversation that isn’t already happening.

There are a variety of reasons for this, but typically it has to do with the way peer review is handled, the way specific disciplines operate, and the way people think research should be done.

Imagine, for example, this fictional conversation between a graduate student and their professor:

If anyone is still with me here, I really encourage you to think further about the significance of Gunkel’s list and the way it has since been used.

It makes me sort of dizzy, actually.

Gunkel basically compiled a very long list of personality-related words.

It has since become essentially canonical for many researchers looking to include a dimension in their research regarding personality.

Gunkel wasn’t a psychologist. He was a brilliant generalist. And yet he has provided something valuable that, presumably, psychology itself—the primary domain for this sort of research—could not provide for Facebook researchers (or perhaps they were simply lazy).

One other aspect of this, related to repetition and peer review, is also important to note.

Once Gunkel’s list was cited, it then became easier for other researchers to cite. I’m sure there’s a name for this phenomenon. It’s a virtuous feedback loop: the more valid Gunkel’s personality list appears, the more valid and useful it becomes.

In his mini-book “What Ideonomy Can Do?” on the MIT website, Gunkel repeatedly emphasizes how ideonomy can clarify thought, broaden potential, and save people from “reinventing the wheel.”

This is how.

Once you have a list, and the list is agreed upon as “valid” by a community, you can—wait for it!—use the list.

Science already does this in some domains.

But in many others, it doesn’t yet know how to have the conversation because there’s no valid canon of ideas.

Besides the relatively straight-forward use of lists as canonical, or comprehensive sources, of knowledge, the real magic will occur in the matching of list items via idea chemistry as I described here. This matching process can actually form a net or matrix around disciplines, providing a desperately needed sense of ground truth defined as the space in which all validated ideas intersect and interface.

Gunkel never claimed that ideonomy’s lists were static, either.

The whole point was that you added and expanded to them over time as new discoveries were made.

He was simply pointing out that if canonical lists become the focus of scientific research, and scientists work harder to agree upon, use, and maintain these lists, the amount of knowledge that can be produced, used, and re-used, is going to be far greater than what is occurring at present—especially with help from artificial intelligence.

This is not going to be an exact science, of course.

Not at first, anyway—because no science is exact at first.

In conclusion, I apologize if the below list of resources doesn’t fit within your interests as a reader of the world’s first and only Patrick Gunkel newsletter. Feel free to skip.

Although my primary goal is raising awareness of Gunkel and his ideas, a secondary objective is sharing information that future scholars may be able to put to good use.

Many of the below papers are included on GGRP’s growing list of academic research papers related to Gunkel. The website also has some papers that are not listed below.

1. Operational Warfare at Sea: Theory and Practice. This October 2008 book by Milan Vego, Professor of Operations at the U.S. Naval War College, cites Gunkel’s list of 638 personality traits, but I did not have access to the specific pages. The reference is made in the context of command/leadership characteristics during naval operations. I would be very interested in seeing the specific pages from this publication if anyone is able to provide them.

2. “Exploring Fairness and Bias in Algorithms and Word Embedding.” In Michael Sosnick’s December 2017 thesis for the University of Pennsylvania’s School of Engineering and Applied Science regarding the measurement of online bias, Sosnick used 590 words from Gunkel’s list of personality traits. In addition, Sosnick relied on Gunkel’s division of these traits into negative, neutral, and positive categories. While Gunkel’s work wasn’t as foundational to the work of other subsequent researchers who used Gunkel’s personality trait list, this list nevertheless played a key supporting role in extending Sosnick’s methodology. The list helped increase the study’s depth and originality in measuring bias. Most unfortunately, Sosnick incorrectly cited Whitman Richards, Gunkel’s MIT advocate, as the source of the list.

3. “Word Embeddings Quantify 100 Years of Gender and Ethnic Stereotypes.” A landmark quantitative study showing how gender and ethnic stereotypes are encoded and have evolved in language over 100 years. The study relied on Gunkel's list of personality traits to provide the foundational resource for the investigation of word embeddings. Literally the study could not have been done without Gunkel’s list. This may have been the first major study to use Gunkel’s list and potentially opened the door for some of the other studies below.

4. Multimodal Composing. This 2019 academic text from the University Press of Colorado “provides strategies for writing center directors and consultants working with writers whose texts are visual, technological, creative, and performative—texts they may be unaccustomed to reading, producing, or tutoring.” In a section intended to help students write electronic portfolios, the book refers to Gunkel’s personality traits list and the fact that it was divided into different categories by stating: “The Ideonomy Project at MIT has a helpful list of 112 positive personality traits that could help a writer brainstorm a specific and focused personality for their Portfolio.”

5. “Engaging Image Captioning via Personality.” This was a 2019 paper produced by Facebook AI Research for the IEEE/CVF Conference on Computer Vision and Pattern Recognition. For this research, Gunkel’s list of 638 personality traits was used to form the basis of an AI recognition model. Without this list, Facebook AI researchers would have been unable to develop a 215-trait schema they used to condition both crowd-sourced captioning and model generation. Neither Gunkel nor the Ideonomy Project is cited in the text, however. The authors state: “We considered 215 possible personality traits which were constructed by selecting a subset from a curated list of 638 traits that we deemed suitable for our captioning task. The traits are categorized into three classes: positive (e.g., sweet, happy, eloquent, humble, perceptive, witty), neutral (e.g., old-fashioned, skeptical, solemn, questioning) and negative (e.g., anxious, childish, critical, fickle).” It is slightly creepy to imagine that the fruits of Gunkel’s effort were gathered and used to power Facebook AI, and yet on the other hand it’s important to recognize that his work did ultimately prove useful to the field.

6. “Enhancing Information Security by Identifying and Embracing Executive Functioning and the Human Behaviors Related to Susceptibility.” This 2021 dissertation by Henry Collier at the University of Colorado is yet another academic article that simply references Gunkel’s list of personality traits without exhibiting any knowledge of (or interest in) ideonomy itself. Hilariously, Collier characterizes Gunkel’s list by stating, “After a thorough investigation into human behaviors, 638 human behavioral traits were identified,” as though Gunkel performed some kind of experiment rather than merely brainstorming with his thesaurus.

7. “Including Human Behaviors into IA Training Assessment: A Better Way Forward!” This June 2022 article, written by Henry Collier of Norwich University, was published in the Proceedings of the 21st European Conference on Cyber Warfare and Security. It cites Gunkel’s list of personality traits but gets his name wrong: “A list of behavioural traits was obtained from an earlier study conducted by MIT's "List Man,"  Peter Gunkel called The Human Kaleidoscope. Gunkel (Gunkel, 1998) developed a list of 638 traits for his study, where he was looking to see if there were limits on human  variability, diversity, and psychogenesis… the extensive list of traits was reduced to a shortlist that could be more practically incorporated into an information security assessment tool.” Gunkel’s list provided a foundational data set to evaluate certain personality types and behaviors for cyber risk potential.

8. “Stigma’s Uneven Decline.” This September 2023 paper, published in the American Sociological Review, used Gunkel’s list as one of several sources to create a “personality traits” stigma dimension in a computational model that measured what sort of language has been associated with disease over time. Kudos to the authors, Best and Arseniev-Koehler, for citing Gunkel correctly—even though their citation had nothing to do with ideonomy.

9. “Personality Understanding of Fictional Characters during Book Reading.” This October 2023 paper—written by a group of Chinese scientists doing cutting-edge Natural Language Processing research—used Gunkel’s personality traits list as the foundational backbone for their entire dataset. They even translated the list into Chinese to build a bilingual dataset that would exhibit consistent personality traits regardless of the original language content. These researchers followed the lead of the Facebook AI Research group listed above. Unfortunately, Gunkel was not cited here, either, as the in-text reference cited only “The MIT Ideonomy Project.”

10. ”Fostering Inclusion among Mixed-Visual Ability Children through Social Robots.” This 2024 PhD thesis in Computer Science and Engineering by Ana Isabel Canciço Neto, University of Lisbon, barely used Gunkel’s personality list to support her thesis, and yet it’s worth commenting on. Her thesis addressed the concept of “social robots” and selected seven emotions from Gunkel’s traits list to help characterize these robots—Happiness, Sadness, Calmnes, Nervousness, Anger, Excitement, and Fury. This citation is remarkable because it demonstrates the afterlife of Gunkel’s list despite the fact that it simply represented a catalog of personality traits. The author of this thesis could have selected these emotions easily “out of thin air,” and yet because we have the perception that science builds on prior knowledge, she required an existing reference to rely upon.