Printer Friendly

Too Much on My Mind: Cognitive Load, Working Memory Capacity, and Framing Effects.

Rational decision theories in economics state that decisions should not be affected by the way a problem or scenario is worded (Diederich & Trueblood, 2018). However, this conjecture has been observed to be false. As Tversky and Kahneman demonstrated in their groundbreaking study in 1981, people are indeed affected by the way a problem or situation is worded. Framing effects refer to the cognitive bias that is attributed to such influences of choice (Tversky & Kahneman, 1981). According to empirical findings, when an option emphasizes positive aspects, and gains (i.e., positive frame), people are more likely to be risk averse. However, when an option focuses on the negative aspects, and losses (i.e., negative frame), people are more likely to be risk-seeking. What is interesting about the tendency to be risk aversive in reaction to positively framed information and risk-seeking in response to negatively framed information is that the expected value of both options is identical (Tversky & Kahneman, 1981; Whitney et al., 2008). Framing effects are believed to be relatively universal, as individuals from varying backgrounds fall prey to this heuristic (Tversky & Kahneman, 1981). It is a reliable effect, but the effect sizes are small to moderate, and the magnitude of the effect depends greatly on the experimental methodology (e.g., between vs. within subjects, type of framing design, such as Gambling-money vs. Asian disease-lives; see meta-analysis by Kuhberger, 1998).

Perhaps, on the account of the universal nature of framing effects, researchers have studied it extensively for the purpose of further understanding the cognitive mechanisms behind decision-making (Gonzalez, Dana, Koshino, & Just, 2005; Igou & Bless, 2007; Kuhberger, 1998; Tversky & Kahneman, 1981; Whitney et al., 2008). Framing effects are present in everyday life, and affect people when making influential decisions in the domain of marketing, advertising and raising social awareness. For example, Jui-Che, Tsai-Feng, and Yi-Chan (2013) found that people were more likely to purchase a "green" product (i.e., the pro-environmental product) when the message was positively framed compared to when the message about the "green" product was negatively framed. Due to the widespread prevalence of decisionmaking, it is important to investigate which cognitive factors interact with the framing of information in order to impact decision-making.

The dual process theory (Evans & Stanovich, 2013; Kahneman, 2011) can be used to explain why people choose to be risk averse or risk-seeking, depending on the frame of the problem and the level of cognitive load. One part of the Dual Process Theory System 1, is responsible for automatic thinking and impulsive decisions. System 1 is also more prone to cognitive heuristics, and under heavy cognitive load, people tend to make more heuristic-based choices compared to reason based choices (Diederich & Trueblood, 2018). In line with this reasoning, there is evidence that under cognitive load, system 1 is engaged, leading to less rational decision making and larger framing effects (Gonzalez, 2005). In terms of explaining framing effects from a neuropsychological perspective, De Martino and colleagues (2006) showed that there were higher levels of amygdala activity in fMRI scans for those who were more susceptible to framing effects. Consistent with the aforementioned finding, activity in the executive, controlled, and inhibitory regions (e.g., the orbital and medial prefrontal cortex) was predictive of people who were less susceptible to framing effects. Taken together, the results of these studies suggest that the automatic and emotional system 1 is a driving factor in risk averse behavior when exposed to positive frames and risk-seeking behavior when exposed to negative frames (De Martino et al., 2006). These results also suggest that under high cognitive load (Gonzalez, 2005) or when the decision-making task evokes an emotionally driven response, framing effects should be larger (Stark, Baldwin, Hertel, & Rothman, 2017).

As mentioned previously, one factor that may impact decisionmaking in framing effect studies is the presence of cognitive load. There is evidence for a negative correlation between cognitive load and decision-making performance on decision-making tasks outside of framing effect domains (Allred, Crawford, Duffy, & Smith, 2016; Gonzalez, 2005; Hinson, Jameson, & Whitney, 2002). For example, Gonzalez (2005) examined the relationship between decision-making performance, innate human cognitive abilities, and task workload (i.e., cognitive load). In this study, participants engaged in a dynamic decision-making task (DDM), in which the task load was manipulated, and each participants' fluid intelligence was measured using the Raven Standard Progressive Matrices Test (RSPMT; Bilker, Hansen, Brensinger, Richard, Gur, & Gur, 2012), which is strongly correlated with working memory capacity (WMC; Wiley, Jarosz, Cushen, & Colflesh, 2011). Gonzalez (2005) found that all participants performed significantly better on the cognitive tasks in the low cognitive load condition than they did on the high cognitive load condition. In addition, those with higher cognitive abilities (according to the RSPMT) performed better on the DDM than those with lower cognitive abilities, in all of the experimental conditions. To summarize, Gonzalez (2005) found that those with lower cognitive abilities were more negatively affected by the cognitive load tasks than those with higher cognitive abilities. Consistent with the previous findings, Hinson and colleagues (2002) reported that participants made poorer decisions (i.e., select options that resulted in smaller gains and larger losses) under conditions of high cognitive load (i.e., maintaining a series of digits). Taken together, the results of the aforementioned studies (Allred et al., 2016; Fletcher, Marks, & Hine, 2011; Gonzalez, 2005; Hinson et al., 2002) provide evidence that cognitive load can impact decision-making. These findings also suggest that individual differences in cognitive abilities should be considered as a moderator factor in decision-making studies.

Because the fluid intelligence task used by Gonzalez (2005) is correlated with measures of working memory, one might predict that in addition to cognitive load, individual differences in working memory capacity (WMC) might also be related to decision-making performance in framing effect studies. Working memory is usually defined as the memory type that incorporates the use of short-term memory, in combination with other cognitive processes to assist in the utilization of short-term memory (Cowan, 2008). Working memory involves the simultaneous storage of information, while also being able to manipulate or work on that information. Working memory also includes an attentional mechanism, which is responsible for managing short-term memory (Cowan, 2008).

There are individual differences in WMC (Conway & Engle, 1994; Gonzalez, 2005; Kyllonen & Christal, 1990). Conway and Engle (1994) showed that those differences were significantly related to retrieval of information from primary memory. Individual differences in WMC also predict performance on a wide variety of cognitive tasks, including decision-making tasks. For example, WMC is positively correlated with performance on reading comprehension tests (Turner & Engle, 1989), on tests of aptitude (e.g., SAT; Mrazek, Smallwood, Franklin, Chin, Baird, & Schooler, 2012), and on tests of fluid intelligence (Fry & Hale, 1996). Support for the connection between WMC and decision-making ability comes from research showing a relatively high correlation between WMC and reasoning ability (i.e., r = 0.80 - 0.90) (Kyllonen & Christal, 1990).

In the realm of decision-making, Cokely and Kelley (2009) found that individuals with higher WMC chose more rationally when faced with risky decisions because they were more likely to compute an expected value instead of relying on heuristic-based strategies. These findings are consistent with previous research showing that individuals who relied on controlled, deliberate cognitive processes (i.e., System 2; Stanovich & West, 2000) were more likely to compute an expected value when making decisions (Frederick, 2005). Stanovich and West (2000) concluded that individuals with lower WMC are limited in their ability to calculate expected values. Further support for the link between WMC and decision-making ability comes from studies showing that people who are low in numeracy, which is the ability to manipulate numbers in the mind, exhibit larger framing effects (Peters & Levin, 2008). Interestingly, numeracy has been found to positively correlate with WMC (Kyttala, Aunio, Lehto, Van Luit & Hautamaki, 2003), suggesting that lower WMC individuals should exhibit larger framing effects. Taken together, the previously described research on WMC, numeracy, and decisionmaking (Cokely & Kelley, 2009; Frederick, 2005; Stanovich & West, 2000) leads to the prediction that higher WMC participants should exhibit smaller framing effects.

Although several researchers have demonstrated that those with higher WMC make more rational decisions (and therefore, should be more resistant to cognitive biases such as framing effects; Cokely & Kelley, 2009; Fletcher et al., 2011), there are some researchers who have found evidence to the contrary within the framing effect domain. For example, Corbin, McElroy, and Black (2010), in their study on working memory and risky choice framing, found that the higher the WMC of an individual, the more likely they were to fall prey to heuristics and biases (i.e., higher WMC individuals had a greater magnitude of framing effects compared to that of lower WMC individuals).

The larger framing effect for higher WMC participants found by Corbin and colleagues (2010) is in line with the Fuzzy-Trace Theory account of framing (Reyna & Brainerd, 1991). According to Fuzzy-Trace Theory, framing effects are the result of attention to gist (simplified information) instead of verbatim or details (non-simplified). In the case of framing, focusing on gist means having a preference for something over nothing in the positive-gain condition vs. preference for nothing over something in the negative-loss condition, while focusing on the exact numbers in the framing problem would reflect attention to verbatim (Reyna, & Brainerd, 1991). Perhaps those with higher WMC are able to simplify complex information more effectively than those with lower WMC. For example, if focusing on gist, one would conclude from a positive frame that some money kept (i.e., the sure option) is better than the gamble option of some money kept or no money kept. Likewise, in the negative frame some money lost (i.e., the sure option) is worse than the gamble option of some money lost or no money lost (Kuhberger & Tanner, 2009; Reyna & Brainerd, 1991). Support for Fuzzy-Trace accounts comes from studies showing that as children age, their WMC increases (Dempster, 1981), and as children age, they tend to exhibit larger framing effects (Reyna & Ellis, 1994; Reyna & Farley, 2006).

Consistent with predictions made by Fuzzy Trace, there is also evidence that higher WMC individuals are more context dependent than lower WMC individuals (Delaney & Sahakyan, 2007). Thus, higher WMC individuals may encode richer contextual information based on elaborative encoding that serves to enable reasoning based on gist. In contrast, lower WMC individuals may pay more attention to numerical information that is prominent and easily available (i.e., verbatim). Based on Fuzzy-Trace, Corbin and colleagues (2010) inferred that because higher WMC individuals in their study showed greater framing effects, they were more reliant on gist compared to lower WMC individuals. Taken together, the previously described research (Corbin et al., 2010; Reyna & Brainerd, 1991) leads to the prediction that higher WMC participants should exhibit larger framing effects. It is important to note that this larger framing effect hypothesis runs contrary to the previously described research on WMC, numeracy, and decision-making (Cokely & Kelley, 2009; Frederick, 2005; Stanovich & West, 2000) which suggests that higher WMC participants should exhibit smaller framing effects. Regardless of the predicted direction of the WMC effect in framing studies, WMC capacity is clearly an important factor to consider within framing studies where cognitive load is induced.

Although working memory capacities vary and WMC clearly relates to decision-making outcomes, previous researchers who studied decisionmaking in framing scenarios did not consider WMC differences (e.g., De Martino et al., 2006; Guo et al., 2017; Igou & Bless, 2006; Whitney et al., 2008). For example, Whitney and colleagues (2008) (using a procedure similar to De Martino et al., 2006) found that a presence of a cognitive load did not increase framing effects. This lack of relationship between the presence of cognitive load and framing effects was inconsistent with previous research showing that cognitive load was related to poorer decision-making (Guo et al., 2017; Hinson et al., 2002). In the cognitive load condition of their experiment, participants were given a sequence of letters that they would be asked about later. Then they were "given" a hypothetical sum of money (i.e., the start value). The task of remembering the sequence of letters while they completed the decision-making task served as the working memory load. In the decision-making phase, participants had to choose between the sure option, (e.g., lose $20), and the gambling option (e.g., 80% chance to keep all $100, or 20% chance to lose all $100). After the participants made their decision between the sure option or gambling option, they were probed to recall the sequence of letters (e.g., "What was the third letter in the sequence?"). Participants in the no load condition did not have to maintain the series of digits while they completed the decisionmaking task. Overall, the proportions of risk averse behavior in the positive frame and the risk seeking behavior in the negative frame were consistent with the general empirical findings (De Martino et al., 2006; Gonzalez et al., 2005; Igou & Bless, 2007; Kuhberger, 1998; Tversky & Kahneman, 1981). However, increasing working memory load actually had no effect on the size of the framing effect. This is because cognitive load reduced the frequency of gambling choices (i.e., increased sure choices) in both framing conditions. The working memory load task did, however, affect the time taken to make the decision, with participants responding faster to the decision-making task in the working memory load conditions, in order to be more accurate in the working memory task. To conclude, cognitive load did not result in greater framing effects, a finding which is inconsistent with dual processing accounts that predict that cognitive load should induce more reliance on the affective heuristic based decision-making component as opposed to the more analytical decision making component.

Instead, Whitney and colleagues' (2008) findings (i.e., a reduction in gambling selections in both framing conditions) are more consistent with the Cognitive-Affective Tradeoff Model proposed by Gonzalez, Dana, Koshino, and Just (2005). According to this model of decision-making, if a particular choice involves costly processing or feels unpleasant, participants will be less likely to choose that option. Thus under high cognitive load (i.e., costly processing) participants experience a reduction in their ability to compute the risk level of the more computational heavy gamble option and instead choose the less computationally effortful sure option. Thus regardless of frame, the Cognitive-Affective Tradeoff account (Gonzalez et al., 2005) suggests that participants under high cognitive load should be more likely to choose the sure option rather than the gamble option. Because Whitney and colleagues (2008) found a significant reduction in the selection of gambling options in both frames under cognitive load compared to no load, their results provide support for the Cognitive-Affective Tradeoff Model. However, it is important to note that although they manipulated cognitive load, Whitney and colleagues (2008) did not specifically examine the potential role of working memory capacity, an individual difference factor that is clearly implicated in decision-making outcomes (Gonzalez, 2005; Stanovich & West, 2000). If participants in their study had higher working memory capacities, they would have been able to better cognitively manage the effects of working memory load, thus masking any effect of working memory load on decision-making. Because WMC was not assessed or reported (via accuracy on the digit span task employed) it is difficult to determine whether cognitive load did not influence the magnitude of the framing effect or whether the framing effect was masked by variance in WMC. One purpose of the current study was to include a measure of working memory capacity as an individual difference factor to determine its relationship to cognitive load and framing effects.

In addition to possible WMC factors, there may be another reason that Whitney and colleagues (2008) did not find an effect of working memory load on framing. Perhaps their cognitive load task was not sufficient enough to influence the magnitude of framing effects. Their working memory load task involved storing a sequence of letters, and processing the framing scenarios presented. Thus their working memory task does meet the standard operational definitions of working memory (i.e., the simultaneous storage of some information while processing other information) (Baddeley & Hitch, 1974; Cowan, 2008; Kyllonen & Christal, 1990). For example, reliable working memory tasks involve solving arithmetic problems and/or grammatical reasoning problems at the same time as one is trying to remember a sequence. It is important to note that although their methodology clearly included load on both the storage (i.e., storing a sequence of letters for later recall) and processing mechanisms of working memory (i.e., decision-making in the framing conditions), Whitney and colleagues' (2008) cognitive load was sequential in nature. Additionally, the decision making task was included as part of the working memory load induction and served as the processing component, instead of sharing the cognitive processing resources with another task. Using a sequential format and the decisionmaking task as the processing component may have been problematic because others (Cowan, 2008; Daneman & Carpenter, 1980) contend that both storage and processing need to be simultaneously engaged as measures of working memory. Thus, the working memory task used by Whitney and colleagues (2008) might not have elicited a sufficiently large enough cognitive load on working memory. Specifically, although their decision-making task did share storage resources (i.e., storage of the sequence while storing the decision making scenarios) it did not have to share any processing resources with another task. The lack of dividing cognitive processing resources in the decision-making task may account for the lack of significance in the potency of framing effects under their high working memory load experimental condition. Increasing the cognitive load by coupling a different working memory task that includes a concurrent storage and processing component along with the decisionmaking processing component might be critical in determining the relationship between working memory capacity, working memory load, and the potency of framing effects.

The purpose of the current study was to investigate framing effect outcomes in terms of ratio or proportion of risk averse and risk-seeking behavior between individuals of lower and higher WMC under conditions of high working memory load. In our experiment, we utilized the operation span task (OSPAN task) (Turner & Engle, 1989) to categorize individuals into two groups-the low WMC group and the high WMC group. We used procedures similar to Whitney and colleagues' (2008) study to determine the effect of cognitive load on the magnitude of framing effects. However, in order to induce greater cognitive strain, we used a different working memory task (i.e., reading span task) that incorporated an additional processing component other than the decision making task.

In terms of our hypothesis regarding the relationship between WMC and the size of the framing effect, three competing outcomes are possible based on the results of previous research. From the perspective of the system 1 and system 2 of the Dual Process Theory (Kahneman, 2011), one might expect to see an increase in the magnitude of framing effects for those with lower WMC, and a decrease in the magnitude for those with higher WMC. The reasoning behind this prediction is that those with lower WMC may not be able to handle the cognitive load as well as those with higher WMC. Therefore, the heuristic driven system 1 will be the principal decision-making system in lower WMC participants, as system 2 will not be able to override the heuristic based system 1. Previous research on dynamic decision-making (e.g., Gonzalez, 2005), supports this prediction, as participants with lower cognitive abilities were affected more by the working memory load task. If the results of dynamic decision-making generalize to the domain of framing effects, lower WMC participants should be even more prone to making heuristic-based decisions, and be more susceptible to framing effects (i.e., exhibit larger framing effects) compared to those with higher WMC, who would be more inclined to make rational decisions (Fletcher et al., 2011) and be less susceptible to framing effects. The prediction that high WMC participants will exhibit smaller framing effects is consistent with Cokely and Kelley's (2009) research on the benefit of high WMC in processing probabilities within framed decision-making tasks.

On the other hand, a second outcome is possible given the results of other decision-making research studies showing that people with higher WMC are more likely to show greater framing effects compared to those with lower WMC, due to the use of a different type of processing and encoding mechanism (Corbin et al., 2010). Based on the Fuzzy-Trace Theory (Reyna & Brainerd, 1991), those with higher WMC may focus more on the gist of the information presented, while those with lower WMC focus more on the verbatim information. Therefore, the results of the current study may be similar to that of Corbin and colleagues (2010) and Fuzzy Trace (i.e., larger framing effects for higher WMC participants). The reasoning behind this second prediction regarding greater framing effects for higher WMC individuals is that under a cognitive load, those with higher WMC may focus more on the gist rather than the verbatim information (i.e., a sure gain in the positive frame is better than a sure loss in the negative frame).

A third and final prediction is also possible given the results of past research (e.g., Whitney et al., 2008). According to the Cognitive-Affective Tradeoff account (Gonzalez et al., 2005), participants under high cognitive load (especially those with lower WMC because they should be less able to handle the cognitive demands) should exhibit smaller framing effects because under costly processing conditions participants are predicted to be more likely to select sure options, which are easier to process (i.e., be risk averse), in both the positive and negative frame conditions.

METHOD

Participants

A total of 60 undergraduates from a small, liberal college in the southeastern United States participated in the study. The final sample size was 42, as participants (n = 18) who were neither high nor low in WMC were excluded from the analysis. The criteria for selecting participants into the low WMC and high WMC groups are outlined later in the preexperimental phase section of the methodology. All the participants were sampled from the psychology research participation pool and participated in exchange for extra credit or course credit. Participants ranged in age from 17 to 23 years, with a mean age of 19.45 years (SD = 1.37). The participant pool was 81.1% female and 17.0% male, and 1.9% other. The ethnic make-up of the participant pool was 84.9% Caucasian, 7.5% African American, 5.7% Hispanic, 1.9% Asian, and 0% other.

Design

The study formed a 2 x 2 x 2 mixed subjects factorial design with working memory capacity (low, high) as the grouping variable, and frame (positive, negative) and starting value (low-$50, high - $100) as the within subjects factors. Frequency of risk aversion (i.e., sure selections) and frequency of risk seeking (i.e., gamble selections) served as the dependent measures for the chi-square analyses. Likelihood to engage in risk-seeking behavior served as the dependent variable for the analysis of variance results. Likelihood to engage in risk-seeking behavior was calculated by dividing the number of times a participant selected the gamble option by the number of decision-making trials within each framing condition. Either chi-square or analysis of variance inferential statistics have been utilized in previous framing effect studies (e.g., De Martino et al., 2006; Guo et al., 2017; Igou & Bless, 2006; Whitney et al., 2008).

Materials

Decision-Making Conditions. All of the stimuli for the decisionmaking conditions are shown in Table 1. Participants were exposed to all four decision-making conditions, where the frame (positive, negative) was crossed with the starting value ($50, $100); 1. Positive-Low Start Value, 2.Positive-High Start Value, 3.Negative-Low Start Value, & 4.Negative-High Start Value). The sure option in the positive frame was phrased in terms of keeping certain amounts ("keep $30", "keep $60"), while the sure option in the negative frame was phrased in terms of losing certain amounts ("lose $20", "lose $40"). It is important to note that for a starting value of $50, keeping $30 in the positive frame is equivalent to losing $20 in the negative frame. Likewise, for a starting value of $100, keeping $60 in the positive frame is equivalent to losing $40 in the negative frame. The gamble options were not identical in all of the decision-making trials, but were kept relatively similar in value to each other (i.e., the gambling option probabilities were around 50-50, but not exactly 50-50). The non-identical gambling options were implemented to ensure engagement and attention in the participants during the decision-making scenario. If all the gambling probabilities were identical, participants may have only considered the sure option in both the frames. Participants were exposed to each of the four decisionmaking conditions, one at a time. The order of exposure to the decisionmaking conditions was randomized for each participant.

Procedure and Tasks

The procedure is depicted in Figure 1. Participants followed the primary procedure outline in Whitney et al. (2008), but with some modifications. First, we measured and created a grouping variable of working memory capacity (via performance on the Operation Span Task). Second, instead of using the digit span task coupled with the decision-making task to induce cognitive load in half of the participants, we used the Reading Span Task (RST) (Daneman & Carpenter, 1980; Engle, Tuholski, Laughlin, & Conway, 1999; Kane, Hambrick, Tuholski, Wilhelm, Payne, & Engle, 2004) to induce working memory load in all of the participants before and while they made their choices in the decision-making phases of the experiment. Third, while under working memory load, participants decided between sure or gamble options across four decision-making conditions instead of 24. Fourth, we utilized a context change phase between each of the four decision-making conditions in order to reduce interference between conditions. Finally, we did not implement stringent time constraints regarding the completion of tasks and presentation of the stimuli; however, we did emphasize to the participants to complete the tasks as quickly and as accurately as possible.

Pre-Experimental Phase: Working Memory Assessment. Prior to the experimental phase of the study, participants' working memory capacity (WMC) was determined via performance on an automated version of the OSPAN task (https://www.wwnorton.com/college/psych/zaps/). The OSPAN is a valid measure of WMC because it requires both processing of some information (i.e., verifying the accuracy of the equations) and storage of other information (i.e., remembering a series of words in order) (Turner & Engle, 1989). Automated versions of OSPAN tasks, such as the one in the current study, correlate well with other measures of WMC and have good internal consistency ([alpha] =.78; Unsworth, Heitz, Schrock, & Engle, 2005). The OSPAN stimuli include 40 mathematical equations and 40, one-to-two syllable, English words. The object of the OSPAN task is to work on some information in memory (i.e., determine the accuracy of mathematical equations), while also maintaining some other information in memory (i.e., remember words in the correct order). Participants were told to accurately determine the correctness of the mathematical equations and to remember as many words in the correct order for a later memory test. Following a sequential presentation paradigm, participants first viewed a math problem (e.g., (10 + 15) / 5 = 6) on the screen, and then indicated whether it was correct or not by clicking "correct" or "incorrect" on the computer screen. Next, participants silently read a single word that appeared (e.g., milk). After 2 seconds, participants repeated the aforementioned tasks on

another mathematical equation and word. Participants practiced two trials and then completed 10 experimental trials, which varied in set size (i.e., three to nine math problems / words). In order to prevent participants from predicting the number of words they would have to memorize, the order of the set sizes was randomized. After each trial, participants completed a recognition memory task; all of the words in the previously shown set appeared on the screen and participants selected the words in the previous set in the order that they were shown. To ensure that participants paid attention to and solved the mathematical equations, we required an 85% accuracy rate on the equations for inclusion of participants' data in the study. The range of possible OSPAN scores was 0 to 40, where higher scores indicated higher WMC.

The OSPAN task was used to categorize participants into two WMC groups using an extreme groups design similar to that reviewed by Conway, Kane, Bunting, Hambrick, Wilhelm, and Engle (2005): the high working memory capacity (higher WMC) group and the low working memory capacity (lower WMC) group. Those with scores in the upper third were selected for the high WMC sample (n = 21), with OSPAN scores ranging from 31 to 40. Those with scores in the lower third were selected for the low WMC sample (n = 21), with OSPAN scores ranging from 9 to 24. Independent samples t-tests revealed that the adopted procedure used to create the extreme groups (i.e., the grouping variable of WMC) worked to establish significant differences in OSPAN scores between the lower and higher WMC groups. The high OSPAN group had significantly greater OSPAN scores (M = 35.62, SD = 3.67) than the low OSPAN group (M = 18.52, SD = 4.02), t(40) = -14.40, p<.001, d = 4.44.

Experimental Phase. During the experimental phase of the study, we first induced a high degree of working memory load by instructing our participants to complete a modified version of the Reading Span Task (RST) (Daneman& Carpenter, 1980; Kane et al., 2004; Unsworth et al., 2005). There were four working memory load induction phases, one for each of the four decision-making conditions. In the RST task, participants completed a reading task while trying to maintain a series of words in their working memory. The RST stimuli were obtained and used with permission from Kane and colleagues (2004). Participants viewed eight sentences on the screen one at a time and verified whether or not the sentences were semantically and syntactically accurate. In a set of eight sentences, three sentences were accurate (e.g., "Although Joe is sarcastic at times, he can also be very sweet") and five were not (e.g., "During the week of final spaghetti, I felt like I was losing my mind). The presentation order of the sentences was randomized. Participants read each sentence out loud and then responded "Yes" to accurate sentences or "No" to inaccurate (nonsensical) sentences. The five nonsensical sentences contain a bolded and underlined word (e.g., spaghetti) that the participants had to remember for a later free recall test after the following decision-making phase. A different set of eight RST sentences was used to induce working memory load across the four decision-making conditions (for a total of 32 sentences). Thus, all participants had to remember the five bolded and underlined words from the nonsensical sentences in the preceding RST while they made their sure or gamble choices during the decision-making phases of the experiment.

There were four decision-making phases of the experiment, one for each of the four decision-making conditions that varied in frame (positive, negative) and starting value ($50, &100). A blocked randomization method was utilized in order to vary the order of decisionmaking condition exposure between participants. In each of the decisionmaking phases, one decision-making scenario appeared on the screen, and the experimenter read out loud the start value, then the sure option, followed the gamble option. Participants made their decision by stating out loud whether they chose the sure or gamble option. Following their decision, the participants were asked to recall the five bolded and underlined words from the directly preceding sentences in the RST in the exact order that they were presented/studied. After they recalled the five words by saying them out loud, all participants completed a context change task (from Sahakyan & Kelley, 2002) designed to induce forgetting in participants and reduce interference from previous trials. For one minute, participants described a room, house, or their favorite store from the moment they entered through the front door. The purpose of the task was to ensure that the cognitive load induced by the RST was controlled (i.e, only applied to one single condition at a time) and to minimize carry-over effects from the previous trial to the next. Upon completion of the context change task, the next trial began (i.e., another RST, choose between sure and gamble options, recall the RST words, and context change). The steps were repeated until participants were run through all four of the decision-making conditions. After the last condition, participants were debriefed.

RESULTS

A 2 x 2 x 2 mixed subjects factorial ANOVA was conducted with WMC (low, high) and the between subjects factor, frame (positive, negative) and starting value (low - $50, high - $100) as the within subjects factors, and risk-seeking likelihood as the dependent measure. Results revealed no significant relationship between WMC group and risk-seeking behavior, F > 1. As shown in Figure 2, there was also no main effect of frame on risk-seeking behavior, F > 1, and no significant interaction between WMC and frame on risk-seeking behavior, F > 1. Thus, participants were not more risk-seeking in response to negatively framed options (M =.56, SD =.48) compared to positively frames options (M =.50, SD =.50) and both low and high WMC participants failed to exhibit significant framing effects.

Despite the lack of framing effects, participants were more risk-seeking in response to decision-making scenarios with high starting values (M =.64, SE =.06) compared to scenarios with low starting values (M =.42, SE =.06), F (1, 40) = 11.32, p =.002 (see Figure 3). This higher level of risk-seeking in response to higher starting value conditions did not vary as a function of frame or WMC as all of the twoway interactions between frame and starting value and between WMC and starting value were not significant, Fs> 1. The three-way interaction between WMC, frame, and starting value was also not significant, Fs> 1. Planned comparisons confirmed the lack of three-way interaction between starting value, WMC, and frame and confirm the presence of main effect of starting value effect across the majority of the WMC by framing conditions. Specifically, participants were significantly more risk-seeking in response to scenarios with high starting values compared to low starting values in the following conditions, with small to medium effect sizes: low WMC participants in the negative frame condition, F 20) = -2.34, p =.03, d = -.59, high WMC participants in the negative frame condition, t(20) = -2.50, p =.02, d = -.49, and in the positive frame condition, t(20) = -2.17, p =.04, d = -.38.

Additionally, a series of chi-square analyses were conducted to determine if the frequency of risk-seeking (i.e., gambling selections) differed significantly from the frequency of risk aversion (i.e., sure selections) within each of six (WMC x Frame x Starting value) conditions. Overall and consistent with the results of the ANOVA, we found that there was no effect of framing on average rates of risk-seeking. However, consistent with the results of the ANOVA, more participants were risk-seeking (i.e., more likely to select the gamble option) in the high start value conditions (M = 64.29%) and more risk averse (i.e., more likely to select the sure option) in the low start value condition (M = 41.67%). The results of the chi-square analyses collapsed across the low and high WMC groups are shown in Table 2, and the results of the chi-square analyses in all six conditions are shown in Table 3.

Positive Frame-Low Starting Value. Overall, when the data were collapsed across high and low WMC groups, fewer participants in the positively framed-low starting value condition chose the gambling option (40.47%) than the sure option (59.52%). However as shown in Table 2, this frequency difference was not significant, [chi square] = 1.52, p = 0.22. As shown in Table 3, this lack of significant frequency difference between selection of the gambling option and selection of the sure option in the positively framed-low starting value condition did not vary as a function of WMC. Fewer low WMC and fewer high WMC participants selected the gamble option than the sure option, however both frequency differences for low WMC and high WMC were not significant.

Positive Frame-High Starting Value. Overall, when the data were collapsed across high and low WMC groups, more participants in the positively framed-high starting value condition chose the gambling option (59.52%) than the sure option (40.48%). However as shown in Table 2, this frequency difference was not significant, [chi square] = 1.52, p = 0.22. As shown in Table 3, this lack of significant frequency difference between selection of the gambling option and selection of the sure option in the positively framed-high starting value condition did not vary as a function of WMC. More low WMC and more high WMC participants selected the gamble option than the sure option, however both frequency differences for low WMC and high WMC were not significant.

Negative Frame-Low Starting Value. Overall, when the data were collapsed across high and low WMC groups, fewer participants in the negatively framed condition-low starting value chose the gambling option (42.86%) than the sure option (57.14%). However as shown in Table 2, this frequency difference was not significant, [chi square] =.86, p = 0.36. As shown in Table 3, this lack of significant frequency difference between selection of the gambling option and selection of the sure option in the negatively framed - low starting value condition did not vary as a function of WMC. Fewer low WMC and fewer high WMC participants selected the gamble option than the sure option, however both frequency differences for low WMC and high WMC were not significant.

Negative Frame-High Starting Value. Overall, when the data were collapsed across high and low WMC groups, significantly more participants in the negatively framed condition-high starting value chose the gambling option (69.05%) than the sure option (30.95%), [chi square] = 6.10, p = 0.014 (see Table 2). As shown in Table 3, this significant frequency difference between selection of the gambling option and selection of the sure option in the negatively framed-high starting value condition did vary as a function of WMC. Significantly more high WMC participants selected the gamble option than the sure option, [chi square] = 3.86, p = 0.05, but for the low WMC participants the higher rate of selecting the gamble option over the sure was not significant, [chi square] = 2.33, p =.13.

Risky Decision-Making: Low and High WMC Comparisons. A series of chi-square analyses were conducted to determine if the frequency of risk-seeking (i.e., gambling selections) differed significantly between high and low WMC participants at each of the four decision-making conditions. As shown in Table 4, overall and consistent with the results of the ANOVA, we found that although there was a trend for more high WMC participants to engage in risk-seeking, there was no significant difference in risk-seeking rates between the high WMC participants (M = 55.95%) and low WMC (M = 50.01%) across any of the conditions, ps> .64.

DISCUSSION

The purpose of the current experiment was to determine if the magnitude of framing effects would vary as a function of working memory capacity under conditions of high working memory load. Results revealed that in general there was no effect of framing on decision-making. Overall, participants in the current study were not more risk averse (i.e., selection of the sure option) in response to financial scenarios that were positively framed in terms of gains (e.g., keep $20). In general, they were also not more risk-seeking (i.e., selection of the gamble option) in response to financial scenarios that were negatively framed in terms of losses (e.g., lose $30). This lack of framing effect was somewhat surprising given that within-subjects manipulations of frame valence used in the current study usually produce larger framing effects than between subject manipulations (see meta-analysis by Kuhberger, 1998). However, consistent with the Cognitive-Affective Tradeoff account, the presence of a high cognitive load may have reduced risk-seeking behavior in both framing conditions, thus eliminating the framing effect.

Although we successfully created extreme WMC groups that differed significantly on working memory capacity scores, we also failed to find a significant relationship between working memory capacity and the magnitude of framing effects. These results are inconsistent with previous research studies that reported either larger framing effects (e.g., Corbin et al., 2010) or more rational decision-making (e.g., Cokely & Kelley, 2009; Fletcher et al., 2011) in those with higher WMC. However, there is weak evidence of a larger partial framing effect in the negative frame for high WMC who were more risk-seeking (than risk aversive) in the negative frame-high start value condition. As shown in the chi-square analyses, this was the only condition where rates of sure selections differed significantly from rates of gamble selections. This larger rate of risk-seeking for high WMC participants is partially consistent with Corbin and colleagues (2010) and Fuzzy-Trace theory (Reyna & Brainerd, 1991). Perhaps high WMC participants in the negative frame-high start value condition, focused more on gist (preference for losing nothing over losing something).

Although we failed to find framing effects, we did find that participants were more risk-seeking in response to decision-making scenarios with high starting values compared to low. This effect of starting value on risk-seeking behavior was stronger in the negative frame condition for high working memory capacity participants. To conclude, we found the highest rates of risk-seeking behavior for high WMC participants in the high starting value, negative frame condition. This finding is consistent with Chiu (2003) who varied the asset level in the framing scenarios, and found that the higher the asset level, the more risk-seeking the participants. For example, modifying Tversky and Kahneman's (1981) Asian Disease problem, they compared 600 people to 6 million people, and in an Asian financial crisis task, they compared $6 million to $6 billion. In both of these framing effect scenarios (i.e., gambling with lives or money), participants were more risk-seeking when the asset level was high. Chiu (2003) stated that a possible reason for this outcome is that higher asset levels create a greater threshold for potential losses when gambling, compared to lower asset levels, and thus increases the likelihood of risk-seeking behavior. In other words, people are more likely to take risks when a potential loss will not hurt them as much. In our current study, when the starting amount (i.e., the hypothetical asset value) was $100, participants may have felt more comfortable with choosing the gamble option as potentially losing $X when they have $100 will not hurt as much potentially losing $X when they have $50.

As discussed previously, we failed to replicate the framing effect in the current study. Reduced framing effects under conditions of high cognitive load are more consistent with the Cognitive-Affective Tradeoff Model proposed by Gonzalez and colleagues (2005), where individuals are predicted to be less likely to choose an option that involves costly processing or feels unpleasurable. Under high cognitive load (i.e., costly processing) participants may experience a reduction in their ability to compute the level of risk associated with the gamble option and instead choose the less cognitively taxing sure option. Thus, the Cognitive-Affective Tradeoff account (Gonzalez et al., 2005) suggests that participants under high cognitive load should be more likely to choose the sure option rather than the gamble option in both framing conditions. Although Whitney and colleagues (2008) found a significant framing effect in both the load and no load condition, they did find a significant reduction in the selection of gambling options in both frames under cognitive load compared to no load, providing support for the Cognitive-Affective Tradeoff Model. Our lack of framing effect could have reflected a greater tendency to select sure options in response to situations of high cognitive load in both framing conditions, but because we did not manipulate load within the experimental design, it is difficult to determine how rates of risk-seeking and risk aversion were altered by the presence or absence of working memory load. Thus, conclusions regarding support for or against the Cognitive-Affective Tradeoff account are limited.

In addition to the role of cognitive load in influencing framing effects, there are several other possible reasons why we failed to find framing effects. First, previous researchers (Fagley & Miller, 1997) have found that framing scenarios within the context of lives (e.g., Asian Disease Problem by Tversky and Kahneman, 1981) result in more risk-seeking than framing scenarios within the context of money or finances, such as the ones used in the current experiment (Fagley & Miller, 1997). This lower rate of risk-seeking in gambling monetary scenarios may explain our lack of significant framing effects.

In addition to the type of framing scenario, the inclusion of a substantial working memory load may have served to reduce the effect of frame overall because it increased the likelihood that participants were selecting options at random, thus performing at chance. Perhaps when working memory is substantially taxed, participants' choices are based more on chance instead of gist or vertamin (Fuzzy Trace), system 1 or system 2 (Dual Process), or a reliance on avoiding unpleasant or cognitively effortful processing (Cognitive-Affective Tradeoff). Consistent with this interpretation, when we look at the overall framing effect (collapsed across WMC conditions and starting value), the likelihood of risk aversion and risk-seeking was around 50%, suggesting that the participants were selecting at random the sure or gamble perhaps without much attention devoted to really understanding the scenarios. A very high working memory load might leave little attentional resources left to engage effectively in the decision-making task that involved substantial processing of information. Consistent with this assertion, Kane and Engle (2000) found that high WMC exhibited reduced ability to suppress interfering information under conditions of high cognitive load. Support for this explanation of the lack of framing effects comes from an examination of the procedure used in the current study. In the experimental phase of the current study, participants first completed the RST task and so they may have focused more on remembering the words presented during the RST (cognitive load) and holding them in memory instead of paying attention to the framing scenarios. They may have also thought that the primary task was remembering the words and that the "filler task" was the decision-making section of the experiment. These factors (i.e., reduced attention and motivation in response to the decisionmaking scenarios) may have contributed to the participants performing at chance level in the decision-making phase, and thus negating any possible framing effects. If Whitney and colleagues (2008) failed to induce sufficient cognitive load, that might explain why they found framing effects. Thus our results may provide evidence that with substantial working memory load, framing effects are reduced or eliminated, although the exact mechanisms guiding the reduction or elimination of the framing effect warrant further investigation.

Another aspect of our procedure may have also inadvertently negated the framing effect in the current study. We did not induce any time pressure for participants to make their decisions. The lack of time pressure may have given a greater opportunity for deliberation and rationalization of the options presented, and reduced the heuristic decision-making outcomes in the current study. Guo and colleagues (2017) found that framing effects are increased under greater pressure to make quick decisions, a finding that is consistent with a dual-process account of framing effects, whereby the quick, intuitive system 1 is engaged to make heuristic based decisions. However, this time pressure interpretation of the results should be considered with caution as Weiguo and Yu (2009) found that time pressure actually reduced the framing effect. Because we did not time how long it took our participants to make their decisions in response to the decision-making scenarios, we have no way of deducing whether or not participants induced time pressure on themselves by making their decisions quickly in order to perform better on the recall task. Future researchers should consider including a timed response procedure in order to determine its effects on the magnitude of framing effects.

Our methodology with regard to the probabilities within the gamble options could have also biased our findings in favor of more risky decisions in response to high starting value scenarios. Specifically, our gamble options were skewed in the positive direction, with all depicting a greater than 50% chance to keep all with a smaller than 50% chance to lose all. According to Whitney and colleagues (2008), scenarios with a 50% or greater chance of a larger gain (in the gamble option) are considered more rational compared the sure option, because one has a greater than 50% chance of finishing with more money than if they had selected the sure option. Indeed, we did find that participants were more risk-seeking across both the positive and negative frames when the starting value was high. It is important to note that our high WMC participants were even more likely to correctly respond in a rational way (to be more risk-seeking) in the negatively framed-high value condition. Perhaps high WMC participants were more likely to interpret the greater likelihood of coming out ahead financially by selecting the gamble option when it was framed in terms of losses and with the higher starting amount, they were subjectively more able to withstand the prospect of a potential loss.

Another possible reason for our inconsistent findings is that we only employed two trials per frame. In order to obtain significant framing effects, more trials with various probabilities represented within the gambling options may be needed. The way the gamble options were phrased could have also contributed to our inability to find framing effects. In a many framing effect studies, the positive gamble is often phrased in the following way, "70% chance to keep all and 30% chance to keep nothing," while the negative gamble is often phrased as, "70% chance to lose nothing and 30% change to lose all." However, in the current study, we replicated the wording employed by Whitney and colleagues (2008) and DeMartino and colleagues (2006) for the positive and negative gamble options such that the gamble was stated in the following way, "70% chance to keep all and 30% chance to lose all" with both keep and lose language employed within each of the framed gambling conditions. Thus, we may have inadvertently reduced the size of the framing effect by not using consistent "keep" language within the positive frame (e.g., keep all, keep nothing) and consistent "lose" language in the negative frame (e.g., lose nothing, lose all). However, it is important to note that both De Martino and colleagues (2006) and Whitney and colleagues (2008) successful found framing effects with the gambling scenarios constructed as we did in the current experiment. Therefore, it is unclear whether the wording of the gambling conditions had a detrimental impact on the size of the framing effect in the current study.

Tversky and Kahneman (1981) suggest that there are different decision weights for gains and losses (i.e., some costs are more influential in decision-making). The weight in terms of reference to costs is steeper in losses than it is for gains (Tversky & Kahneman, 1981). Therefore our starting amounts may have been too high in the high starting value conditions. Additionally, according to Tversky and Kahneman (1981) specific percentages used in the gambling options can affect decision-making. Thus the percentages we employed in our Positive Frame-High Starting Value condition (i.e., the "80% chance to keep all $100") may have been too large and influenced participants to choose the gamble option, even though the selection of the sure option was the predicted outcome.

To summarize, given the aforementioned limitations of the current experiment, one may strengthen the design of this experiment by adding more decision-making conditions to create more variance, as well as a control condition that has no cognitive load in order to address the direct relationship between the cognitive load and decision-making. It would also be beneficial to include consistent "keep" and "loss" wording within the positive and negative gambling option conditions, respectively. Because cognitive load might have been too high, future research might include use a shorter version of the RST task to induce working memory load. A modification of the instructions to stress the importance of both the RST task and the decision-making task as well as collection of reaction time to the decision-making trials may also be warranted in future research. Another potential direction of this research could be to utilize neuroimaging technology similar to research by De Martino and colleagues (2006) determine if a cognitive load is present and determine which brain areas are activated in individuals with different working memory capacities. It might also be interesting to vary the type of framing scenario to determine the relationship between WMC, load, and framing effects using the Asian Disease problem. Finally, future researchers should employ larger sample sizes in order to increase power to detect significant results.

Although we failed to replicate the framing effect, the current results are important to report for several reasons. First, our results suggest a wealth of possible constraints on framing effects in the presence of cognitive load. Consistent with the Cognitive-Affective Tradeoff account, cognitive load may have served to eliminate framing effects in the current study. Next, we did replicate the effect of a high starting value and our results provide additional support that high starting values within decision-making scenarios result in greater risk-seeking behavior. Finally, we did find partial support for a role of WMC in framing effects. Consistent with Fuzzy Trace Theory (Reyna & Brainerd, 1991), high WMC participants were more risk-seeking in response to negatively framed scenarios with a high starting value, suggesting that they attempted to reduce interference under a high working memory load by focusing on gist instead of verbatim, mathematical calculations. The current results run contrary to previous research on decision-making and imply that in some situations (and in some experimental designs), high

WMC participants may be more likely to be more risk-seeking in response to negatively framed information with a high starting value under situations of high cognitive load. The current results suggest that WMC may related to framing effect variance, but future research is warranted to determine the exact nature of this relationship and under what cognitive conditions such evidence of the relationship surfaces. Because cognitive factors can apply to most individuals in the general public and have an influence on how decisions are made in multiple disciplines such as marketing, psychology, and even judicial cases, it is important to conduct more research in order to understand the relationship between the cognitive factors of WMC, working memory load, and framing effects.

REFERENCES

Allred, S. R., Crawford, L. E., Duffy, S., & Smith, J. (2016). Working memory and spatial judgments: Cognitive load increases the central tendency bias. Psychonomic Bulletin & Review, 23(6), 1825-1831.

Baddeley, A. D., & Hitch, G. (1974). Working Memory. In G. H. Bower (Ed.), Psychology of Learning and Motivation (Vol. 8, pp. 47-89). New York: Academic Press.

Bilker, W. B., Hansen, J. A., Brensinger, C. M., Richard, J., Gur, R. E., &Gur, R. C. (2012). Development of abbreviated nine-item forms of the Raven's Standard Progressive Matrices Test. Assessment, 19(3), 354-369.

Chiu, Y.--C. (2003). Is there a framing effect? The asset effect in decisionmaking under risk.Chinese Journal of Psychology, 45(2), 171-182.

Cokely, E. T., & Kelley, C. M. (2009). Cognitive abilities and superior decision making under risk: A protocol analysis and process model evaluation. Judgment and Decision Making, 4(1), 20-33.

Conway, A. A., & Engle, R. W. (1994). Working memory and retrieval: A resource-dependent inhibition model. Journal of Experimental Psychology: General,123(4), 354-373.

Conway, A. R. A., Kane, M. J., Bunting, M. F., Hambrick, D. Z., Wilhelm, O., & Engle, R. W. (2005). Working memory span tasks: A methodological review and user's guide. Psychonomic Bulletin & Review, 12(5), 769-786.

Corbin, J., McElroy, T., & Black, C. (2010). Memory reflected in our decisions: Higher working memory capacity predicts greater bias in risky choice. Judgment and Decision Making, 5(2), 110-115.

Cowan, N. (2008). What are the differences between long-term, short-term, and working memory? Progress in Brain Research, 169, 323-338.

Daneman, M., & Carpenter, P. A. (1980).Individual differences in working memory and reading. Journal of Verbal Learning & Verbal Behavior, 19(4), 450-466.

Delaney P.F., & Sahakyan L. (2007). Unexpected costs of high working memory capacity following directed forgetting and contextual change manipulations. Memory & Cognition, 35(5), 1074-1082.

Dempster, F. N. (1981). Memory span: Sources of individual and developmental differences. Psychological Bulletin, 89(1), 63-100.

De Martino, B., Kumaran, D., Seymour, B., & Dolan, R. J. (2006). Frames, biases, and rational decision-making in the human brain. Science, 313(5787), 684-687.

Diederich, A., & Trueblood, J. S. (2018). A dynamic dual process model of risky decision making. Psychological Review, 125(2), 270-292.

Engle, R. W., Tuholski, S. W., Laughlin, J. E., & Conway, A. R. A. (1999). Working memory, Short-term memory, and general fluid intelligence: A latent-variable approach. Journal of Experimental Psychology: General, 128(3), 309-331.

Evans, J. S. B. T., & Stanovich, K. E. (2013). Dual-process theories of higher cognition: Advancing the debate. Perspectives on Psychological Science, 8(3), 223-241.

Fagley, N. S., & Miller, P. M. (1997). Framing effects and arenas of choice: Your money or your life? Organizational Behavior and Human Decision Processes, 71(3), 355-373.

Fletcher, J. M., Marks, A. D., & Hine, D. W. (2011). Working memory capacity and cognitive styles in decision-making. Personality and Individual Differences, 50(7), 1136-1141.

Frederick, S. (2005). Cognitive reflection and decision making. Journal of Economic Perspectives, 19, 24-42.

Fry, A. F., & Hale, S. (1996). Processing speed, working memory, and fluid intelligence: Evidence for a developmental cascade. Psychological Science (0956-7976), 7(4), 237-241.

Gonzalez, C. (2005). Task workload and cognitive abilities in dynamic decision making. Human Factors, 47(1), 92-101.

Gonzalez, C., Dana, J., Koshino, H., & Just, M. (2005). The framing effect and risky decisions: Examining cognitive functions with fMRI. Journal of Economic Psychology, 26(1), 1-20.

Guo, L., Trueblood, J. S., &Diederich, A. (2017). Thinking fast increases framing effects in risky decision making. Psychological Science, 28(4), 530-543.

Hinson, J. M., Jameson, T. L., & Whitney, P. (2002). Somatic markers, working memory, and decision making. Cognitive, Affective & Behavioral Neuroscience, 2(4), 341-353.

Igou, E. R., & Bless, H. (2007). On undesirable consequences of thinking: framing effects as a function of substantive processing. Journal of Behavioral Decision Making, 20(2), 125-142.

Jui-Che, T., Tsai-Feng, K., & Yi-Chan, T. (2013). Influences of framing effect and green message on advertising effect. Social Behavior & Personality: An International Journal, 41(7), 1083-1098.

Kahneman, D. (2011). Thinking, fast and slow. New York, NY, US: Farrar, Straus and Giroux.

Kane, M.J., & Engle, R.W. (2000). WM capacity, proactive interference, and divided attention:

Limits on long-term memory retrieval. Journal of Experimental Psychology: Learning, Memory, and Cognition, 26, 336-358.

Kane, M. J., Hambrick, D. Z., Tuholski, S. W., Wilhelm, O., Payne, T. W., & Engle, R. W. (2004). The generality of working memory capacity: A latent variable approach to verbal and visuospatial memory span and reasoning. Journal of Experimental Psychology: General, 133(2), 189-217.

Kuhberger, A. (1998). The influence of framing on risky decisions: A metaanalysis. Organizational Behavior & Human Decision Processes, 75(1), 2355.

Kuhberger, A., & Tanner, C. (2010). Risky choice framing: Task versions and a comparison of prospect theory and fuzzy-trace theory. Journal of Behavioral Decision Making, 23(3), 314-329. https://doi.org/10.1002/bdm.656

Kyttala, M., Aunio, P., Lehto, J. E., Van Luit, J., & Hautamaki, J. (2003).Visuospatial working memory and early numeracy. Educational and Child Psychology, 20(3), 65-76.

Kyllonen, P. C., & Christal, R. E. (1990). Reasoning ability is (little more than) working-memory capacity?!. Intelligence, 14(4), 389-433.

Mrazek, M. D., Smallwood, J., Franklin, M. S., Chin, J. M., Baird, B., & Schooler, J. W. (2012). The role of mind-wandering in measurements of general aptitude. Journal of Experimental Psychology: General, 141(4), 788798.

Peters, E., & Levin, I. P. (2008). Dissecting the risky-choice framing effect: Numeracy as an individual-difference factor in weighting risky and riskless options. Judgment and Decision Making, 3(6), 435-448.

Reyna, V. F., & Brainerd, C. J. (1991). Fuzzy-trace theory and framing effects in choice: Gist, extraction, truncation, and conversion. Journal of Behavioral Decision Making, 4(4), 249-262.

Reyna, V. F., & Ellis, S. C. (1994). Fuzzy-trace theory and framing effects in children's risky decision making. Psychological Science, 5(5), 275-279.

Reyna, V. F., & Farley, F. (2006). Risk and rationality in adolescent decision making: Implications for theory, practice, and public policy. Psychological Science in the Public Interest, 7(1), 1-44.

Sahakyan, L., & Kelley, C. M. (2002).A contextual change account of the directed forgetting effect. Journal of Experimental Psychology: Learning, Memory, and Cognition, 28(6), 1064-1072.

Stanovich, K., & West, R. (2000). Individual differences in reasoning: Implications for the rationality debate? Behavioral and Brain Sciences,23(5), 645-665.

Stark, E., Baldwin, A. S., Hertel, A. W., & Rothman, A. J. (2017). The role of rational and experiential processing in influencing the framing effect. The Journal of Social Psychology, 157(3), 308-321.

Turner, M. L., & Engle, R. W. (1989). Is working memory capacity task dependent? Journal of Memory and Language, 28(2), 127-154.

Tversky, A., & Kahneman, D. (1981). The framing of decisions and the psychology of choice. Science (New York, N.Y.), 211(4481), 453-458.

Weiguo, H., & Yu, H. (2009). Time stress affecting the framing effect of risk decision-making. Psychological Science (China), 32(3), 694-696.

Whitney, P., Rinehart, C. A., & Hinson, J. M. (2008). Framing effects under cognitive load: the role of working memory in risky decisions. Psychonomic Bulletin & Review, 15(6), 1179-1184.

Wiley, J., Jarosz, A. F., Cushen, P. J., & Colflesh, G. J. H. (2011). New rule use drives the relation between working memory capacity and Raven's Advanced Progressive Matrices. Journal of Experimental Psychology. Learning, Memory, And Cognition, 37(1), 256-263.

Author Note: Medhini Urs, Department of Psychology, Florida Southern College; Leilani B. Goodmon, Department of Psychology, Florida Southern College; Jordan Martin, Department of Psychology, Florida Southern College. The authors would like to thank Samantha Burnsed for helping with data collection.

The authors declare no potential conflicts of interest with respect to the research, authorship, and or publication of this article.

Medhini Urs, Leilani B. Goodmon, & Jordan Martin

Florida Southern College

Author info: Correspondence should be sent to: Dr. Leilani B. Goodmon, Ph.D., Department of Psychology. 111 Lake Hollingsworth Drive Lakeland, FL 33801-5698 Email: lgoodmonriley@flsouthern.edu

Caption: FIGURE 1. Procedure events in the experimental phase of the study. After each decision-making condition, participants provided their choice (sure or gamble) verbally and then verbally recalled the 5 words from the RST portion in the order in which they were studied/presented.
TABLE 1. The Four Decision-Making Conditions
Starting
Value                       Positive Frame     Negative Frame

Amount
               Sure             Keep $30           Lose $20
Low ($50)                     70% chance         65% chance
               Gamble       to keep all $50    to keep all $50

                              30% chance         35% chance
                            to lose all $50"   to lose all $50

               Sure             Keep $60           Lose $40
High ($100)                   80% chance         75% chance
               Gamble       to keep all $100   to keep all $100

                              20% chance        25% chance to
                            to lose all $100    lose all $100

TABLE 2. Chi-square comparisons of the proportion of risk seeking
and risk aversive decision-making choices as a function of frame
(positive, negative) and starting value (low--$50, high--$100)
collapsed across working memory capacity (low, high). * significant
at the .05 level.

Starting                        Positive           Negative
Value                            Frame              Frame

                 Gamble          40.47%             42.85%
Low ($50)         Sure           59.52%             57.14%
               Chi-Square      1.52 (.22)        0.86 (0.36)
                 (sig.)

High ($100)      Gamble          59.52%             69.05%
                  Sure           40.47%             30.95%
               Chi-Square      1.52 (.22)         6.10*(.01)
                 (sig.)

TABLE 3. Chi-square comparisons of the proportion of risk seeking
and risk aversive decision-making choices as a function of frame
(positive, negative), working memory capacity (low, high), and
starting value (low - $50, high--$100). * significant at the .05
level.
                            Positive Frame
Starting                          Low                High
Value                             WMC                WMC

                 Gamble          38.09%             42.86%
Low               Sure           61.90%             57.14%
($50)          Chi-Square         1.19               0.43
                 (sig.)          (0.28)             (0.51)

                 Gamble          57.14%             61.90%
High              Sure           42.86%             38.09%
($100)         Chi-Square         0.43               1.19
                 (sig.)          (0.51)             (0.28)

                             Negative Frame
Starting                          Low                High
Value                             WMC                WMC

                 Gamble          38.09%             47.62%
Low               Sure           61.90%             52.38%
($50)          Chi-Square         1.19               0.04
                 (sig.)          (0.28)             (0.83)

                 Gamble          66.67%             71.43%
High              Sure           33.33%             28.57%
($100)         Chi-Square         2.33              3.86*
                 (sig.)          (0.13)            (0.049)

TABLE 4. Chi-square comparisons of the frequency of risk seeking
decision-making choices between low and high WMC groups as a
function of frame (positive, negative) and starting value
(low--$50, high - $100).
                            Positive Frame
Starting                          Low                High
Value                             WMC                WMC

                 Gamble          38.10%             42.86%
Low                n               8                  9
($50)          Chi-Square     0.06 (0.81)
                 (sig.)

High             Gamble          57.14%             61.90%
($100)             n               12                 13
               Chi-Square     0.04 (0.84)
                 (sig.)

                             Negative Frame
Starting                          Low                High
Value                             WMC                WMC

                 Gamble          38.10%             47.62%
Low                n               8                  10
($50)          Chi-Square     0.22 (0.64)
                 (sig.)

High             Gamble          66.68%             71.42%
($100)             n               14                 15
               Chi-Square     0.04 (0.85)
                 (sig.)
Note--percentages reported in terms of frequency of choosing the
gamble option out of the total sample size within each WMC group
(n = 21). The chi-square results are based on the number of
participants within each condition who chose the gamble option.

FIGURE 3. Average likelihood of risk seeking as a
function of working memory capacity--WMC (low,
high), frame (positive, negative), and
starting value (low, high)

                              Low WMC
                    Low Starting   High Starting
                        Value        Value

Positive Frame            0.38     0.57
Negative Frame            0.38     0.67

                             High WMC

Positive Frame            0.43     0.62
Negative Frame            0.48     0.71

Note: Table made from bar graph.

FIGURE 2. Average proportion of gamble
choices as a function of frame
and working memory capacity (WMC).

                  Low WMC   High WMC
Positive Frame      0.48     0.52
Negative Frame      0.52     0.60

Note: Table made from bar graph.


Please Note: Illustration(s) are not available due to copyright restrictions.
COPYRIGHT 2019 North American Journal of Psychology
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2019 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Urs, Medhini; Goodmon, Leilani B.; Martin, Jordan
Publication:North American Journal of Psychology
Date:Dec 1, 2019
Words:11009
Previous Article:Is Avoiding Meat Prosocial? Field Research on Meat and Vegan Influence on Prosocial Behavior.
Next Article:Do Not Pet the Crawfish: Starting an Invertebrate Behavioral Lab.
Topics:

Terms of use | Privacy policy | Copyright © 2020 Farlex, Inc. | Feedback | For webmasters