Neural Correlates of Math Anxiety and Ability on Price Promotions and Consumer Decisions

• ¹ Drexel University, School of Biomedical Engineering, Science, and Health Systems, United States
• ² LeBow College of Business, Drexel University, United States
• ³ University of Pennsylvania, Department of Family and Community Health, United States
• ⁴ Children's Hospital of Philadelphia, Division of General Pediatrics, United States

Previous research suggests that math anxiety not only increases consumers’ tendencies to make computational errors but also influences their ability to make the best numerical decision (Suri, Monroe, & Koc, 2012). However, other research has attributed such differences in people’s performance on lack of numerical ability rather than math anxiety (Peters et al., 2006). Hence, the question remains as to which one of these two aspects of human personality is more likely to impact consumers decisions involving numbers like when choosing between competing price promotions. This research examines the neural correlates and behavioral performance of consumers as they make judgments about price promotions. Using both behavioral and neural measures this research will provide a deeper understanding of the effect of math anxiety and numerical ability on day-to-day computation decisions. In this study, we aimed to identify the relationship between math anxiety and ability on consumer choices using fNIRS (Ayaz et al., 2013). We also observed their task performance under low and high cognitive loads. Our intention is also to investigate if fNIRS can capture pre-frontal activation changes when people with different levels of math anxiety evaluate price promotions (Ashcraft & Kirk, 2001). We believe this is the first attempt to incorporate a neurophysiological method to assess computation performance differences associated with anxiety and ability in an ecologically valid manner when comparing price promotions. Eighteen participants between the ages of 19 and 43 (8 females, mean age = 23 years) volunteered for the study. All participants completed a different survey before the experiment day. This survey included the sMARS (short Mathematics Anxiety Rating Scale) (Alexander & Martray., 1989) and the ETS mathematical ability test. The sMARS is a short version of MARS (Richardson & Suinn, 1972), which evaluates the level of apprehension and anxiety of people when they confront a situation that includes mathematics or mathematical calculations. The protocol consisted of two blocks (no load and with load) and each block consisted of 18 trials. Participants completed two practice trials before starting each block. Price (in dollar form) and rental days as a multiplier were given as variables. On the center of the display, below the variable information, two discount options were presented. First option was in dollar ($) form and the second one was in percentage (%) form. Participants were asked to select the option, which gives the highest discount value or the lowest total price. On block 2 (with load condition), each trial started with a display of five alphanumeric characters on the screen for 4000 milliseconds. Participants were asked to memorize these alphanumeric characters during the discount task. After completion of each discount trial, three different sets of alphanumeric characters were shown on the computer screen. Participants were asked to recognize the set with the same characters that they were shown at the beginning of the trial. Behavioral efficiency for the load and no-load conditions can be seen in table 1. We compared two variables: math anxiety levels (F1,16 =.13, p < 0.718) and math ability levels (F1,16 =0.06, p < 0.8). None of the results were significant. The difference in prefrontal cortex hemodynamics for math anxiety levels are summarized in fig 2 for the load and no-load conditions. For the left (F1,16 =0.98, p < 0.337), and right (F1,16 =0.01, p < 0.948) hemispheres there was no significant difference overall. Next, we compared the difference in math ability levels. The prefrontal cortex hemodynamics results are summarized in fig 3. for the low and high math ability levels. fNIRS based oxygenated hemoglobin changes had a significant effect in the left hemisphere between the low and high math ability groups (F1,16 =9.92, p < 0.006). fNIRS based oxygenated hemoglobin changes also showed a significant effect in the right hemisphere between the low and high math ability groups (F1,16 =5.42, p < 0.033). Behavioral results indicated that participants with low math anxiety had a higher behavioral efficiency (accuracy/response rate) score in both the load and no-load conditions compared to the participants with a high math anxiety score, as expected (Ashcraft & Kirk, 2001). Prefrontal cortex hemodynamics results showed that participants with lower levels of math anxiety have higher levels of brain activation in the left hemisphere where the more math anxious participants had higher levels of activation in the right hemisphere. For math ability, we see that participants with high ability for math had higher levels of activation in both right and left hemispheres compared to those with lower math ability. In the behavioral performance, the more proficient group scored better in both conditions than the low ability group. In conclusion, this study demonstrated it is possible to detect math anxiety levels and math ability correlates in brain activation using fNIRS. Fig. 1 (Table 1). Behavioral Efficiency. Comparison of math anxiety levels shows higher math anxious participants had lower efficiency in the no load and load conditions. Similarly, comparison of math ability suggests the participants who are less proficient in math scored lower in both conditions as well. Fig. 2 Comparison of math anxiety levels for both conditions. Prefrontal cortex oxygenation changes show under the no load condition and load condition that less math anxious participants had higher levels of oxygenated hemoglobin in the left hemisphere (left) where more math anxious participants had higher levels of oxygenated hemoglobin in the right hemisphere. The whiskers are standard error of the mean (SEM). Fig. 3. Comparison of math proficiency for both conditions. Prefrontal cortex oxygenation changes show under both no load and load conditions the high ability group had higher brain activation in both the left hemisphere (left) and right hemisphere. The whiskers are standard error of the mean (SEM).