Session 1 (12:30-1:30)

Do students learn better with material that is perceptually hard to process? While evidence is mixed, recent claims suggest that placing materials in Sans Forgetica, a perceptually difficult-to-process typeface, has positive impacts on student learning. Given the weak evidence for other similar perceptual disfluency effects, we examined the mnemonic effects of Sans Forgetica more closely in comparison to other learning strategies across three preregistered experiments. In Experiment 1, participants studied weakly related cue-target pairs with targets presented in either Sans Forgetica or with missing letters (e.g., cue: G_RL, the generation effect). Cued recall performance showed a robust effect of generation, but no Sans Forgetica memory benefit. In Experiment 2, participants read an educational passage about ground water with select sentences presented in either Sans Forgetica typeface, yellow pre-highlighting, or unmodified. Cued recall for select words was better for pre-highlighted information than an unmodified pure reading condition. Critically, presenting sentences in Sans Forgetica did not elevate cued recall compared to an unmodified pure reading condition or a pre-highlighted condition. In Experiment 3, individuals did not have better discriminability for Sans Forgetica relative to a fluent condition in an old-new recognition test. Our findings suggest that Sans Forgetica really is forgettable.

Barry Sopher, Jacqueline Gelman (presenting)

Maximizing the Growth Rate of Wealth vs Maximizing Expected Utility of Wealth: The St. Petersburg Gamble Revisited

Some remarkable recent work (Peters, 2011) has demonstrated that Bernoulli’s conjectured solution to the “St. Petersburg Paradox”—to maximize the expected value of the natural log of wealth—is the exact solution to a completely different problem—to maximize the time averaged growth rate of one’s wealth. This “time resolution of the St. Petersburg paradox” can be understood as highlighting more generally the need to take account of whether the wealth generation process is ergodic or not. In particular, if we provisionally accept the hypothesis that the growth rate of one’s wealth is a suitable optimand for an economic agent to focus upon, then the transformation of wealth represented by what is conventionally called the utility function takes on a new meaning: it is the transformation of wealth whose stochastic behavior is ergodic—meaning that both the “ensemble” or “cross sectional” distribution is the same as the “time average” distribution. Notably, this transformation depends not upon individual risk preferences, but rather upon the physical environment. For example, the St. Petersburg Gamble, which has a geometric distribution over time, requires the natural log transformation to yield an ergodic process, while a simple binomial gamble (win a dollar or win 0 with equally probability in each round, for example) requires no transformation of wealth—it is already ergodic. So the appropriate function to optimize with differs, for the same individual, across different gambles. We report on an experiment designed to exploit the implications of Peters’ theory. We compare gambling behavior in a truncated (to 10 periods) version of the St. Petersburg Gamble to behavior in a 10 period Binomial Gamble. We conceive of the experiment as a horse race between risk-neutral expected utility and the growth rate maximization hypothesis. These two hypotheses yield distinct predictions for the threshold of the cost of playing the gamble above which a decision maker should not be willing to play the St Petersburg Gamble, but rather similar predictions for the two theories for the Binomial Gamble. Our results are, briefly, that the vast majority of subjects are remarkably consistent, behaving as if they have a distinct threshold in mind for each type of gamble, for a given initial (endowed) wealth level that they have to work with, and from which they pay the cost of playing he gamble, or which they keep if they do not play the gamble. The threshold levels increase systematically with the initial wealth level, but the implied transformation of wealth (or utility function) ends up being too variable to be consistent with either hypothesis. Further, overall behavior in the St. Petersburg gamble is generally closer to risk neutral behavior than is behavior in the Binomial Gamble. A third alternative hypothesis that the initial wealth subjects are endowed with amounts to “house money” and is not integrated into wealth, is explored. Overall, the full implications of the multiplicative nature of wealth dynamics in the St. Petersburg Gamble does not seem to impinge on decision makers’ thinking. A future experiment in which subjects have decision aids to give them feedback on hypothetical decisions they might make before deciding to gamble or not is discussed.

Aversive learning normally induces alterations in sensory function as the brain’s sensory systems are tuned to optimize detection and discrimination of threat-predictive stimuli. Anxiety disorders can disrupt behavioral discrimination between threat-predictive and neutral stimuli, resulting in overgeneralization of negative affective responses to non-threatening situations. We thus hypothesized that anxiety could disrupt learning-induced improvement in sensory discrimination. We tested perceptual discrimination between similar odors before and after aversive conditioning in which one odor co-occurred with a mild electrical shock and one occurred without the shock. Participants exhibiting normal levels of trait anxiety developed a larger skin conductance response to the shock-predictive odor, along with substantial improvement in their perceptual discrimination between the two odors that was absent in odor only and shock only controls. Repeated exposure to the odors without shock partially extinguished the skin conductance responses but the perceptual effect persisted. By contrast, the subset of participants with high levels of trait anxiety developed comparably sized skin conductance responses to both odors, suggesting overgeneralization of their conditioned response and did not show any perceptual improvement following conditioning. Learning-induced perceptual plasticity, a normal part of healthy sensory processing, can thus be impaired in people with high levels of trait anxiety.

The nonverbal social cues that accompany speech (e.g., facial expressions, gestures, and eye gaze) can be as communicatively significant as the content of the speech itself. In this paper, I identify what I argue is a very common phenomenon: we tend to allocate nonverbal cues in a way that’s sensitive to conversational participants’ levels of respective social power such that people with more power receive comparatively more positive and affirming nonverbal cues than people with less power. I call this ‘nonverbal marginalization’. In section one, I introduce and empirically situate nonverbal marginalization and in section two I argue that we can understand the harms created by nonverbal marginalization through the lens of the philosophical literature on epistemic injustice. I go on in section 3 to argue that the nonverbal marginalization concept can contribute to our understanding of two important phenomena in social psychology: imposter syndrome and achievement gaps. I conclude with an empirically and theoretically informed discussion of how we can go about undermining the harmful effects of nonverbal marginalization.

The current study normalizes 43 picture sets, showing stereotypical and random events, based on their degree in difficulty to order (given the individual pictures in a random sequence) and level of scriptedness. The Rating Stimulus Task assigned participants (N=60) 15 of the total 43 picture sets with the individual pictures in randomized order. Participants rated the picture sets on their levels of difficulty in ordering and factors scriptedness. Participants named the picture sets for the second study, the Itemization Task. Difficulty and scriptedness were highly correlated. Difficulty and scriptedness were found to be continuous variables rather than binary. The Itemization Task assigned participants (N=58) 15 of the 34 titles chosen to represent one or more of the picture sets. Participants listed 3 to 20 action associated with the picture set title. Scriptedness was correlated with longer average listing of actions. Frequently listed actions were combined to create one list of associated actions for each picture set title, considering transition probability for each action. Picture sets were evaluated based on how their “ground truth” actions matched the list of actions made by participants. Future directions are encouraged to utilize similar normalizing studies and further investigate the non-binary nature of scripted events

This study investigates the prosody of college students with autism spectrum disorder (ASD). Five NT and 5 ASD college students were given the PEPS-C (to assess pragmatics). Syntax was assessed using a forced-choice comprehension test, and prosody was assessed by having participants say pragmatically neutral sentences. ASD participants did significantly worse than NT participants on the production section of the PEPS-C, but not on the comprehension section. They did almost perfectly on the syntactic comprehension test. In terms of prosody, ASD participants spoke more slowly than NT participants, and unlike NT participants who had a mid-sentence amplitude peak, ASD participants’ sentences decreased monotonically. The F0 of all NT participants increased monotonically during the first half of the sentence and then decreased monotonically. In contrast, there was a great deal of variability in F0 among the ASD participants, with one decreasing monotonically, one having a sentence-final peak, one having 2 peaks and one having 3 peaks. In sum, even ASD participants with normal receptive prosody have aberrant expressive prosody for pragmatically neutral sentences, with there being substantial individual differences.

A central canon in theory of mind (ToM) research is that three-year-olds fail standard false-belief tasks, whereas four-year-olds pass. The Theory of Mind Mechanism (ToMM) theory (Leslie, 1994) posits that change in inhibitory power (IP) accounts for this difference. A recent computational implementation of this theory (Wang, Hemmer, & Leslie, 2019) quantified the developmental role of IP using group aggregated data. Established findings summarize group-aggregated data; but computational processes and developmental change take place within individuals, not in an ‘aggregate brain’. The only study of developmental change within individuals is Baker, Leslie, Gallistel, and Hood (2016) who, in a series of longitudinal single-case change-point studies, found that false-belief performance remains unstable for long periods. Here we extend the computational ToMM model to the unique Baker et al. (2016) data. We find that, rather than age, IP is a powerful, and plausibly the main, contributor to developmental change in ToM reasoning.