Emotion Detection and the Mimetic Faculty

On March 6, 2019, the Guardian reported how emotion detection technology—which usually refers to a form of facial recognition that can differentiate between a range of expressed emotions—had rapidly emerged as a $20 billion-dollar industry.

The emotion detection industry is guided, in part, by claims of Rosalind Picard, the author of the influential book Affective Computing and co-founder of the MIT Media Lab-affiliated startup Affectiva.[1] Computers that can “read” human emotion, Picard suggests, will be able to react more effectively to human desires, creating responsive digital interfaces through the ability to “see” emotion and mirror it accordingly. Affective interfaces, Picard proposes, could even act as a means of therapy, training people to more effectively express or interpret emotion. When it comes to the massive tech companies investing in these technologies, companies like Amazon, Facebook, and Microsoft, the ideal of a mirrored emotional interface is compounded by political economy. Detecting emotions will, potentially, allow these companies to manipulate human desire—directing consumption and attention—prior to conscious awareness, linked with a supposedly innate human capacity to mirror the emotions of others, be those others people or computers.

These claims about emotion detection are similar to a concept Walter Benjamin sketched in several short essays from the early 1930s. In these writings, Benjamin proposed something he termed “the mimetic faculty,” or the ability to copy and to generate similarities, which he claimed was one of the greatest competences of human life.[2] But this capacity does not remain unchanged over time. There’s something about modernity that has caused this mimetic faculty to wane, to fall into disuse, Benjamin suggests. Perhaps this is one way we can understand emotion recognition—that, in the loss of the capacity to generate and identify similarity, we’ve developed machines to perform this task, and delegated mimesis to them.

This assumes that the face is the primary medium for mimesis. While there are forms of emotion detection that analyze the voice, for instance, the majority of which work to see emotion as displayed on faces. Benjamin, too, appears to suggest a primacy to the face—which is one of the reasons he often made positive reference to physiognomy, that famed pseudoscience that reads character from countenance. In his “Little History of Photography,” Benjamin contemplated the photographic portraits collected in August Sander’s 1929 book, Face of Our Time. This book provides a “tremendous physiognomic gallery” and, Benjamin claims, “is more than a picture book. It is a training manual” (520). Is this because Sander allows us to see—through the photographic halting of motion, visualizing what Benjamin called “the optical unconscious”—similarities that were once sensed but now lost, and can be regained through sufficient technological training?

Perhaps. But to end there would be to overlook how Benjamin almost always links mimesis with the cosmic, usually by way of astrology. The deferral to the face is a reduction of the mimetic faculty, not its most idealized form. In the fragment “On Astrology,” Benjamin writes that the modern waning of the mimetic faculty is represented in how “modern man…really only recognizes facial similarities, and no longer has much ability to recognize bodily similarities” (685). Mimesis provided by physiognomy and facial expression is limited, and, Benjamin suggests, astrology is a faded remnant of a past time in which humanity was able to sense its mimetic relations with the cosmos beyond the face. “The horoscope must above all be understood as an originary totality that astrological interpretation merely subjects to analysis” (685). Echoing Benjamin’s theorization of jetztzeit, the astrologer makes “the conjunction of two stars; it must be grasped in an instant” (696).

This indicates Benjamin’s mimetic faculty may be radically different than the kinds of similarity implied by emotion detection. Astrology produces similarity in the act of interpretation, leading to flashes of insight in the charting of a constellation. Or, mimesis is not mirroring, and the deferral to the face—and the idea that mimesis is merely a mirroring of the face of another—is indication of the mimetic faculty’s loss.

Theodor Adorno, while living in Los Angeles, wrote several essays denouncing astrology and the occult, claiming they were—like much Adorno wrote about—expressions of popular authoritarianism and a creeping, fascist conformity perpetuated by the culture industry.[3] I’d suggest that the sense of mimesis offered by emotion recognition follows Adorno. The similarities exist to standardize forms of relation, conditioning emotion, determining action.

But Benjamin’s turn to astrology is quite different than Adorno’s. Instead, it suggests a kind of mimesis that is dialectical and rupturing, one that does not find similarities, but makes similarities where they may not otherwise exist. Or, one of the places we can locate Benjamin’s legacy in digital culture is in striving for a kind of mimesis that is not merely about the mirroring of faces and the movement of emotion from a body to a machine. Rather, Benjamin’s mimetic faculty works to generate similarities in difference, to create links out of separations, and to refuse to let all be colonized by the same. It suggests a mimesis that cannot be found in systems of emotion recognition.

[1] Rosalind Picard, Affective Computing (Cambridge, MA: MIT Press, 1997).

[2] Walter Benjamin, Selected Writings, Volume 2, edited by Michael W. Jennings, Howard Eiland, and Gary Smith (Cambridge, MA: Harvard University Press, 1999), 694. Henceforth cited in the text.

[3] Theodor Adorno, The Stars Down to Earth and Other Essays on the Irrational in Culture, edited by Stephen Crook (London: Routledge, 1994).

Add new comment

Log in or register to add a comment.