Two academic papers
One thing that I haven’t done in this blog is point readers to my “professional” publications. In my day job I am an academic who teaches and does research in the areas of habit formation and heuristics. Here are two recent articles that relate to these two topics.
First is an article that looks at the phenomenon of response chunking. Let’s take a step back.
When one observes animal behavior, one typically sees behavior that makes sense, is rational (i.e., functional), and perhaps in some sense “optimal.” A bird flies to a patch, forages, and departs. A mouse encounters a blocked route, and digs its way under the obstacle. In each case, though, we should appreciate that the animal — the bird or the mouse – selects its moment-to-moment behavior from a broad range of potential behaviors. Why does the bird forage and not groom? Why does the mouse dig and not attack? Or stated broadly, how does a coherent set of behaviors emerge from the repertoire of all possible behaviors?
Such a question might sound preposterous, until one appreciates syndromes in which action selection is impaired. Individuals with compulsions and stereotypies or patients with Parkinson’s disease, for instance, all exhibit syndromes that at least partially relate to difficulties surrounding action selection. How movements become chunked into coherent actions; how actions are represented as goal-oriented plans; and how actions are flexibly selected by the immediate environment remain open questions for behavioral science.
So, this paper looks at just one of these: the chunking of single responses into higher-order response categories. The paper uses pigeons pecking at lit keys in an operant chamber. The term “VI schedule” in the title just refers to a particular patterning of rewards in relation to the pigeons’ pecking behavior.
The second paper has to do with how people (not pigeons!) perceive patterns in outcomes, even when these patterns are not actually present. In this case, the paper is looking at something termed “recency bias.” This refers to how people will sometimes use the repeated occurrence of something as a predictor of the next occurrence being more or less likely.
For example, let’s consider the recency bias known as the “gambler’s fallacy.” This is a bias in which subjects treat a recurring outcome as less likely to occur in the future, despite “knowing” that the outcome is the result of a flat probability. Put colloquially, subjects act as if the outcome has temporarily “used up its luck.” For example, if a coin is tossed 10 times, and generates 7 heads, many subjects will report that heads should be less likely to occur. What my students and I have been attempting to get a handle on are the psychological processes that underpin the gambler’s fallacy bias. In various independent study projects, we’ve manipulated outcome salience, gain-loss contexts, learned helplessness, and impulsivity.
Traditionally, the gambler’s fallacy result has been interpreted as the result of subjects utilizing a choice heuristic known as the “law of small numbers” (Tversky and Kahneman, 1971). This law says that subjects act as if segments of any length of a random sequence must reflect the true proportion. In other words, the law of small numbers says that people tend to exaggerate the degree to which a small sample will resemble the parent population. Whether 1,000 tosses of a coin or 10, subjects will assume that a fair coin will produce an equal number of heads and tails. So, after the run T, H, H, H, subjects will report that the next coin toss is much more likely to come up T(ails) – the gambler’s fallacy result.
My own hypothesis for understanding the gambler’s fallacy choice bias, though, originated from my appreciation of foraging models. I termed this hypothesis the “replenishment hypothesis.” I’ve illustrated the essential features of the replenishment hypothesis in the figure above. Essentially, the hypothesis posits that humans are acting like foraging animals when they speculate as to outcome frequencies. When one has been away from a patch (e.g., tails), then that patch is more likely to contain food, than another patch (e.g., heads) that one has been depleting. That is, each patch replenishes at some rate, akin to a tank that is filling with water. A “sample” is akin to drawing some amount of water from a particular tank. The figure above might be akin to T, H, H, H, and the level of each take would correspond to the perceived likelihood that either T or H would occur on the next trial.
The paper: Parsing negative and positive recency biases in terms of top- down and bottom-up processing, uses ideas drawn from how perceptual illusions are often deconstructed to examine negative and positive recency biases.