Big Data, Big Deal!

Selectivity is the source of meaning.

Posted Dec 26, 2018

To think is to forget differences, generalize, make abstractions. ~ J. L. Borges, Funes el memorioso

We live in the era of big data. The corporations that have the data, Google, Amazon, Facebook et al., use them to their advantage. They often know what you want before you do, and they are daily getting better at this. Google et al. experiment with the settings on their sites and let the data tell them what works best (for their bottom line, that is). Their approach to the numbers game is not science, but a sort of fine-tuning of technology; perhaps it can be called applied science, or, as the phrase goes ‘data science.’ When there are only data, there is no need for perspective or theory or inference. The data are – to use the fogeyish plural – clear. Uncertainty has been conquered and the dreaded standard error, or ‘margin of error,’ has been marginalized into extinction. In this brave new world, where data define reality, it is a quaint memory that I spent so many hours blogging against the idea of free will (Krueger, 2018a). The data now show that there is none, right? Oh well, it seemed important at the time.

¡Un momentito por favor!

I am not being entirely serious. I still think a case can be made for the idea that the data cannot tell all. For one thing, they cannot make meaning. It takes a human cortex hooked up to a limbic system to do that. We do not have to look farther than the classic social psychological studies on human influence. Be it conformity (Asch, 1956), obedience (Milgram, 1963), or bystander behavior (Darley & Latané, 1968), the data – small as they were – condensed themselves into percentages. Take Asch’s result that in about 1/3 of the trials the human subject grew a spine and spoke truth to a confederated majority of liars. Let the data be big and it is still ca. 33% conformity. If Mr. Data looked at the data, he would report the result as 33% conformity. If you asked him what he makes of it, he would be at a loss. Social psychologists, the students they teach, and the interested public, have been amazed and appalled by this figure because they expected and demanded that there should be no conformity at all. Perceptual truth, that is, when judging the lengths of lines as in Asch’s experiment, must trump social consensus, that is, when the hired confederates call out a line different in length as being identical in length. This expectation gives meaning to the result, a meaning, which is not inherent in the numbers. So, are people weak? Are they driven by a herd instinct like dumb bovines? What can be done to make people rational, independent, and truthful?

If we had a theory that humans are social imitation machines, that they have evolved an instinct of imitation, which has stood them in good stead for 10,000 generations, then we’d be amazed and appalled that on 2/3 of the trials the respondents broke rank. We’d ask what is wrong with these people? Why would they flirt with social breakdown for the sake of a silly line (Krueger & Massey, 2009)? If the data cannot create meaning, if theory and human inference remain relevant, there is hope that science can avoid becoming a playground for bean counters. Data will continue to interface with theory and inference, and we humans can continue to participate in the construction of models of how our world works.

A recent brouhaha may further illustrate that this is, in fact, a fairly deep point to contemplate (Chater et al., 2018). With my colleagues Teppo Felin & Jan Koenderink – who did most of the work – I co-authored an article questioning the wisdom of evaluating human perception and cognition against a normative model treating omniscience as not only possible but as desirable. We termed this view the ‘all-seeing-eye’ approach (Felin, Koenderink, & Krueger, 2017). We argued – to brutally simplify our point – that humans, like other animals, have expectations and interests that render most of the incoming stimuli irrelevant. This led us to see the famous Gorilla-in-the-basketball-court experiment in a different light, as it were. Simons & Chabris (1999) reported that of the subjects instructed to count the number of passes made by one team, 46% did not notice the person in a gorilla suit strutting across the court. Following the all-seeing-eye paradigm, the authors emphasized the minority of subjects who did not notice gorilla man. Alternatively, we marveled at how many subjects were able to focus on the task they were given, and ignore that which was defined as irrelevant.

The traditional view rides on the appeal of the folk belief that blindness is bad and that seeing more can only be better. Yet, there is an equally compelling folk belief that we must be able to concentrate and not be distracted by noise or intrusion. Have we come to a theoretical stalemate where the choice between seeing all and seeing only what is task relevant is a matter of taste? I don’t think this would be the right conclusion. Notice that there is an asymmetry: The all-seeing-eye position assumes that anything that is sufficiently ‘salient’ should be noticed, where salience is determined ad hoc by the experimenters or other observers not engaged in the focal task. In contrast, the focal-attention position says that everything that does not directly bear on the task at hand should be screened out. By this definition, being blind to that which does not matter is a success. The asymmetry is that only the focal-attention position is clear a priori about what should be noticed.

I think that our failure to suppress perceiving unwanted content is a far greater adaptive liability than failing to notice that a co-worker had a haircut (i.e., change blindness). The big data companies know that this is our weakness and they are eager to exploit it.

Gino 2:58

J. Krueger
What SHOULD I do?
Source: J. Krueger

After discussing Francesca Gino’s Rebel Talent (Krueger, 2018b), I watched her business keynote on the LAVIN site. Gino explains that “Rebels fight this natural tendency of avoiding conflict, of avoiding tension, and instead they embrace it.” Rebels do not ask the natural question “What should I do?” but they ask “What could I do?” [1] The critical segment starts 2:45 minutes. At 2:58 minutes, Gino shows a slide with the question “What SHOULD I do?” There is no other substantive content on this slide, which makes it a good slide as slides go. However, this slide also shows about 30 faded corporate logos from Pepsico (upper left corner) to Bank of America (lower right; see screenshot on the left). Gino does not comment, and it’s the only slide in her presentation with intrusive advertisements. Would it not be nice to not see them? Ironically, and this is not my main point here, Gino’s telling us here what we should do and not what we could do if we want to be rebels. Don’t rebel against HER advice!

Why do LAVIN, Gino, and Pepsico submit us to these logos? I suppose they do because they can, which raises the question of why not all slides come with gentle reminders of what to drink and where to bank. Years ago, I wondered in the privacy of my own mind if I couldn’t boost my professor’s income by bringing ads into the classroom or by offering the front and back of my shirt as ad space. I dismissed these ideas as impractical and dystopian, but perhaps their time has come. Was Gino being rebellious when floating her ad-studded slide? At any rate – and the all-seeing-eye be damned – there is not enough blindness in the world. 

Funes el memorioso

It is easily forgotten that our ability to forget is our memory’s blessing. An all-recalling-mind would be debilitating. Why ask of vision what we couldn’t bear in memory?

[1] Gino - correctly, I suppose - assumes that the audience considers the 'should' question to be the natural one. Presenting the 'could' question as the rebellious alternative, she seizes the element of surprise, setting the stage to refute the audience's mistaken beliefs. As Davis (1971) would put it: "That's interesting!" 

References

Asch, S. E. (1956). Studies of independence and conformity: I. A minority of one against a unanimous majority. Psychological Monograph, 70(9, whole number 416).

Chater, N., Felin, T., Funder, D., Gigerenzer, G., Koenderink, J., Krueger, J. I., Noble, D., Nordli, S., Oaksford, M., Schwartz, B., Stanovich, K., & Todd, P. (2018). Mind and rationality: An interdisciplinary debate. Psychonomic Bulletin & Review, 25, 793-826.

Darley, J. M., & Latané, B. (1968). Bystander intervention in emergencies: Diffusion of responsibility. Journal of Personality and Social Psychology, 8, 377-383.

Davis, M. S. (1971). That's interesting! Philosophy of Social Science, 1, 309-344

Felin, T., Koenderink, J., & Krueger, J. I. (2017). Rationality, perception, and the all-seeing eye. Psychonomic Bulletin & Review, 24, 1040-1059.

Krueger, J. I. (2018a). Five arguments for free will. Psychology Today Online. https://www.psychologytoday.com/us/blog/one-among-many/201803/five-arguments-free-will

Krueger, J. I. (2018b). Controlled eccentricity. Psychology Today Online. https://www.psychologytoday.com/us/blog/one-among-many/201812/punctuated-rebellion

Krueger, J. I., & Massey, A. L. (2009). A rational reconstruction of misbehavior. Social Cognition, 27, 785-810.

Milgram, S. (1963). Behavioral study of obedience. Journal of Abnormal and Social Psychology, 67, 371-378.

Simons, D. J., & Chabris, C. F. (1999). Gorilla in our midst: Sustained inattentional blindness for dynamic events. Perception, 28, 1059– 1074.