The read-write memory of the human brain allows for the storage of inputs through symbols. Like a Turing machine, our deep neural network system can read in a “code” and store its information in an infinite number of finite patterns. Just as a Turing machine has an infinite storage tape but a finite number of symbols (0’s and 1’s) to represent an input, our neural networks use a finite set of objects to represent an infinite number of possibilities. This means that we can acquire and store knowledge of an unlimited number of objects – be it the abstract emotional concept of “love” or a concrete object such as a farm; we may represent these in our minds as a heart symbol and the image of a barn and field, respectively. However, it is clear that these concepts entail so much more in reality than their stored symbolic representations. This simplification of inputs into a symbolic representation, discussed heavily in my Symbolic Systems course, was also investigated and subsequently analyzed in both LINGUIST 1 Readings, including Gary Lupyan’s “The paradox of the universal triangle: Concepts, language, and prototypes” and John Russell Rickford’s “Unequal partnership: Sociolinguistics and the African American speech community.”
Lupyan first describes the structure of perceptual representation with the example of storing the memory of a dog, as “a very simple neural network model in which a verbal cue (“dog”) is associated over time with various dogs [from previous experiences],” that develops to become more “progressively dissociated with features of the experiences that vary arbitrarily and associated with features that are most diagnostic of the category,” such as with images of the most common 2-eared, 4-legged dogs we tend to see every day (Lupyan 4). Using various experimental studies, Lupyan tests to see if this standard specific-to-abstract progression of memory holds true for other paradigms of constrained symbolic variants, such as the triangle. Using triangles as a microdomain of constrained symbolic variants, Lupyan’s investigation revolved around the prototypical representation of a “triangle” as a standard horizontally-oriented equilateral or isosceles triangle in the human mind. Firstly, he noticed that when asked to specifically “Draw a triangle” rather than to “Draw a three-sided polygon,” participants tended to draw “more canonical triangles” (Lupyan 8). He also found that when prompting participants to rate the typicality of these triangles from 1-7 from most unusual to most typical. In continuation of previous trends, participants “judged triangles drawn to the triangle prompt as more typical than triangles drawn to the three-sided polygon prompt,” proving that at the most fundamental neural level, “category names help to form a kind of idealized perceptual state—a prototype of sorts,” rather than a broad categorical representation. This conclusion was confirmed in a second experiment where speeded cue-picture verification was conducted, asserting that even categorically constrained cues, such as “triangle”, elicits a trend towards a particular object from that category due to a clear prototypical representation.
Lupyan’s findings impact far more than the way our minds represent images of dogs and triangles. The fact that we are innately biased towards stored prototypical representations means that we continue to find the “atypical” representation unusual. This perceptual marginalization of the atypical often leads to subconscious prejudices – just as we tend to find minorities groups such as women and people of color “atypical” and by extrapolation “less suitable” for positions in various industries, such as in government, business, and STEM fields. Likewise, it is the reason we often stereotype certain linguistic variants, be it the “refined and educated” English seen in those speaking Received Pronunciation, aka “The Queen’s English” or the “redneck and yokel” English of the Southern United States. These biases are unjust, unwarranted, and inappropriately prejudiced, yet they continue to perpetuate due to the universality of stored prototypical representation. This consequence is undoubtedly existent in response to African American Vernacular (AAVE), as discussed by Rickford. While AAVE involves many standardized rules such as copula absence or word-final consonant cluster simplification, just as other non-biased language dialects have structured developmental rules, AAVE is unfairly negatively stereotyped against in our media, economy, culture, and education fields. Even the field of sociolinguistics necessary to understand these biases is impacted – take, for example, the fact that, “Despite more than a quarter century of concentrated work on AAVE, only a handful of African American faculty of any specialization exist in linguistics” (Rickford 168-169). Lupyan’s findings are clearly causally linked to Rickford’s consequential analysis and evidently causes diversity in our society to be, albeit perhaps often unintentionally, marginalized and underrepresented. This discovery raises several crucial questions in my mind: Are we able to actively override the tendency towards the prototypical to better engage and include diversity in our society? If so, how should we can we go about doing this to ensure the long-term coexistence of all minorities, now and in the future?
The answers to these questions are surely equal parts enigmatic and important, and likely play a major role in studies of prejudices of race, sex, and ethnicity. And while we may not currently know these answers, we must all play a vital role in taking a first step forward by being actively aware of these unwarranted assumptions and consciously being self-aware to ensure inclusivity, even at our most subconscious perceptual level.
In a sense, I am optimistic, for we are not simply Turing machines. We differentiate ourselves y reacting to our surrounding environments; we respond to ubiquitous external stimuli every day, be it through our everyday conversations or representation in the media. If we collectively work to adjust these stimuli to better include “atypical” representation, these representations will surely become more mentally “typical” and thus inputted into our own read-write memory as the norm. Let us work together to be the creators of our own stimuli, and thus the creators of a more inclusive “mind” and more accepting society.
No comments:
Post a Comment