In The Dark (Senior Thesis)
In The Dark is a series of interactive data sculptures visualising how bare plural generics, a class of generalisations, can conceal information and lead readers astray. Below are some highlights but please read my full thesis if you would like to learn more and don’t hesitate to reach out if you have any questions!
Special thanks to Susan Gelman, Sarah-Jane Leslie and Andrei Cimpian for their research on generic language that has inspired this project, and to Arts at Michigan for their funding support.
This project has since been awarded 1st Place in the 2021-2022 Blue Multi-Term Award, one of the annual Pamela J. Mackintosh Undergraduate Research Awards open to all undergraduates at the University of Michigan, and the Integrative Project Written Thesis Award, awarded to one student in the graduating class within the Art&Design major.
Artist Statement
Humans generalise for two main reasons: to organise information efficiently and to enable inductive inferences for decision making. While some generalisations can be useful, others can distort reality and contribute to harmful stereotyping.
Each double-walled porcelain sculpture confronts the viewer with visualised statistics for a generalisation that is expressed without a quantifier. From a distance, only the outer form - the impression of 100% prevalence - is visible. When the viewer approaches the sculpture, a motion detector turns on a light inside and the inner volume - the actual prevalence - is revealed through the translucent porcelain as a casted shadow. Without physically closing the distance, the viewer is left in the dark.
Every sculpture in the series looks the same from the outside but have drastically different casted shadows, prompting the viewer to question why the generalisations all come across as true despite having drastically different prevalences.
Part 1: Prevalence Spectrum
In The Dark is intended to be viewed as two complimentary parts: Part 1 provides context for the interactive experience in Part 2 to be discussed.
Spanning across the walls is a spectrum comparing six generalisations and their statistical prevalences. The generalisations I have chosen fall under what linguists classify as bare plural generics : unquantified statements (do not possess quantifiers such as “all”, “most” or “some”) that express generalisations but do not carry information about how many members of the category have the given property. As a result, generics are ambiguous and very tolerant of exceptions - this is precisely what the prevelance spectrum aims to visualise.
*Please note that all figures are US-based and updated as of March 2022 for installation purposes
Each generic is visualised with a corresponding Euler diagram sculpture where the coloured area represents its prevalence. For example, “Plastic waste is recycled”:
Exact percentages are omitted from the physical installation to encourage the viewer to focus on the relative size differences between the coloured area on each sculpture rather than on the numbers.
[<1%] “Bacteria are harmful”
Most bacteria are harmless — less than 1% of the different types actually cause disease in humans (AHRQ). In fact, many are helpful and some are even essential for human health. For example, gut bacteria play a crucial role in maintaining our immune system and protect against pathogens (Thursby et al.); we need good bacteria to fight off the bad ones.
[8.7%] “Plastic waste is recycled”
According to latest government figures, only 8.7% of plastic waste was recycled in 2018 in the United States (EPA). In 40 years, less than 10% of plastic waste has ever been recycled in the United States (Sullivan).
See also Plastic Wars, a PBS documentary that investigates the future of plastics, tracing back to the plastics industry’s role in promoting recycling in the 1980s, allegedly to sell more plastics.
[<50%] “Ducks lay eggs”
Only adult, fertile and female ducks lay eggs — this is less than 50% of all ducks. Yet “ducks lay eggs” strike us as true while “ducks are female” does not, even though there are more female ducks than there are egg-laying ducks. (Leslie)
[37%] “The general public says it is safe to eat GM foods”
In 2015, the Pew Research Center conducted a study that compared opinions of the American public and scientists on a range of scientific and social matters. The starkest opinion difference was on whether it is safe to eat genetically modified foods: 37% of the general public compared with 88% of scientists (Funk).
[88%] “Scientists say it is safe to eat GM foods”
[100%] “Diamonds are made of carbon”
Diamond is a substance that is defined by its chemical makeup: a mineral made of pure carbon (Britannica). Therefore, all diamonds are necessarily made of carbon.
Even though the generics above all come across as true, their statistical prevalences range from 1% all the way to 100%. Among generics accepted as true, not only is there massive variance in statistical prevalence, many are often accepted despite weak evidence. This suggests that majority prevalence is not necessary for a generic to be judged as true - a very puzzling feature of generics.
Every sculpture has a unique crystalline glaze that each straddle a fine line of beauty and decay. The iridescent finish alludes to gemstones while the irregular crystals remind the viewer of petri dishes and disease. Much like this dual impression of crystalline surfaces, the prevalence spectrum seeks to be descriptive and open-ended, leaving it up to the viewer to decide whether they want to continue endorsing certain generics after learning about their statistical prevalences.
Part 2: Biomorphic Generics
Part 1 confronts the viewer with a puzzling feature of generics: whether a generic is accepted as true has an almost negligible relationship with its statistical prevalence. Part 2 encourages the viewer to consider the implications of said feature through three motion-activated lighted sculptures.
From a distance, only the outer form — the impression of 100% prevalence — is visible. When the viewer approaches the sculpture, a motion detector turns on a light inside and the inner volume — the actual prevalence — is revealed through the translucent porcelain as a casted shadow. Without physically closing the distance, the viewer is left in the dark. The lighted sculptures share the same titles and data as the first half of the prevalence spectrum. While both are data sculptures, the Euler diagram format conveys the data directly whereas the motion-activated format illustrates how proximity affects whether the same data is accessible.
“Bacteria are harmful”
“Plastic waste is recycled”
“Ducks lay eggs”
Building off the prevalence spectrum, the lighted sculptures are Euler diagrams reimagined as three-dimensional forms where the outer volume represents 100% and the casted shadow represents the actual prevalence. Each sculpture is labelled with the corresponding generic it represents. The three lighted sculptures share generics with the first half of the prevalence spectrum: “Bacteria are harmful”, “Plastic waste is recycled” and “Ducks lay eggs”. I chose to reimagine them in particular because they do not satisfy majority prevalence - above 50% - thus deviate most starkly from the widespread impression implied by the generic statements. It is interesting to consider why most of us would not be comfortable claiming that “bacteria are harmless” even after learning that 99% of them are in fact harmless.
Much like how generics sound broadly true before close inspection, the white porcelain sculptures look essentially the same at a distance, symbolising how subconscious generics we hold are taken to be similarly true, neutral and unassuming. As the viewer approaches the sculptures, their neared physical presence activates motion-detection lights and the sculptures are lit from within, casting shadows of different sizes that correspond with the generic’s data. Borrowing from the idiom “in the dark”, the light-based data reveal symbolises the viewer’s escape from ignorance, enabled by their willingness to interact at a close proximity.
Part 1: Prevalence Spectrum highlights a disconnect between a generics’ truth valuation and its statistical prevalence. Part 2: Biomorphic Generics identifies the role of proximity in cementing ignorance. Piecing together the two parts, the viewer is encouraged to critically think about why generics come across as true despite having drastically different prevalences, how this can pose epistemic and political threats, and examine their own use of generic language.
