Decision‐making competence and cognitive abilities: Which abilities matter?

Funding information Marianne and Marcus Wallenberg Foundation, Grant/Award Number: 2014-0173 Abstract Decision-making competence is a skill that is associated with numerous positive life outcomes. Even though multiple cognitive abilities have been shown to predict decision-making competence, few studies have incorporated a large test battery tapping into several cognitive abilities concurrently in the same models. The current paper presents a study that sought to investigate which cognitive abilities predicted overall decision-making competence in adults using hierarchical regression analysis. A cognitive test battery, comprising abilities such as general intelligence, executive functions, numeracy, visuospatial ability, and time perception, was administered to 182 participants. Results indicate that both general intelligence, which was consistently the strongest predictor, and numeracy contributed independently to overall decision-making competence. Executive functions did predict overall decision-making competence, while all predictors were included in the models. A novel finding concerns the relationship between time perception and decision-making competence. The complementary roles of these cognitive abilities are discussed.


| INTRODUCTION
Decision making is a pervasive feature of everyday life encompassing both trivial activities, such as deciding what to eat for lunch, and more pressing and existential issues with potentially detrimental consequences, such as whether to undergo a risky surgical procedure or to invest in the stock market. Initial models of human decision making have traditionally viewed human decision outcomes as products of rational, cognitive processing where the probability and the utility of each alternative are clearly and exhaustively calculated. However, pioneering research has shown that individuals consistently fall prey to various behavioral biases, such as the framing effect (Tversky & Kahneman, 1981) and sunk costs (Arkes & Blumer, 1985), that runs counter to the notion that humans adhere to traditional normative theories of decision making.
Mounting evidence suggests that there are stable individual differences in how susceptible people are to various decision-making biases and how well people adhere to normative principles (Bruine de Bruin et al., 2007;Parker et al., 2018). A comprehensive multifaceted test has been developed called the "Adult Decision-Making Competence"(A-DMC) test (Bruine de Bruin et al., 2007), which consists of six subcomponents that together capture an overall estimate of "decision-making competence." The A-DMC has been demonstrated to be related to normative outcomes such as socioeconomic status, less maladaptive and risky behavior in youth, antisocial behavior, substance abuse, and adolescent delinquency (Parker et al., 2018). Therefore, the value of knowing what individual and social circumstances give rise to greater decision-making skills cannot be overstated. Decision-making competence has been linked to several individual characteristics, such as personality variables  and decision-making styles (Dewberry et al., 2013), as well as individual cognitive abilities. Performance on the A-DMC is intricately tied to individual cognitive abilities, where studies have shown correlations with general intelligence (Bruine de Bruin et al., 2007), executive functions (EFs; Del Missier et al., 2012), numeracy (Del Missier et al., 2012), and memory capacity (Del Missier et al., 2013). Even though a multitude of cognitive abilities has been shown to predict decision-making competence, few studies have incorporated a large test battery tapping into several cognitive abilities concurrently in the same models. Given that these cognitive abilities, as well as the neural mechanisms subserving them, are overlapping and interlinked, it is essential to investigate models where multiple abilities are taken into account. Otherwise, there is a risk that any study investigating the influence of a single cognitive ability (e.g., general intelligence) may fail to reveal a genuine effect of a third and related factor (e.g., working memory [WM]) that correlates with the ability used in the analysis. In addition, the superior decision making is likely supported by a complex orchestration of several cognitive abilities, thus warranting further investigation of these putative associations. The current paper presents a study that sought to investigate which cognitive abilities predicted overall decision-making competence in adults using hierarchical regression analysis. Below, we provide an overview of previous studies that have looked into the relationship between various cognitive abilities and decision-making competence. First, we will present previous research on DMC and the relation with intelligence and EF, after which we provide an overview of previous research on DMC and spatial and temporal processing.
Last, we will highlight research on various numerical abilities and their role in DMC (see Table 1 for an overview of prior research reporting correlations between DMC and cognitive abilities).

| Decision-making competence, intelligence, and EFs
General cognitive abilities, such as EFs and intelligence, have consistently been linked to various aspects of decision-making competence.
One conventional way of assessing general intelligence is by administering Raven's Progressive Matrices (RPM;Raven, 1976). General intelligence-or "g"-can be conceptualized as the apex of a hierarchy of several underlying abilities (Carroll, 1993), where "fluid" intelligence and "crystallized" intelligence are primary components that constitute g (Cattell, 1963;Horn, 1968). Fluid intelligence refers to the ability to reason about and solve novel problems that require an understanding of new rules or concepts, while crystallized intelligence denotes the ability to reason using previous experience and learned concepts.
Intelligence is prima facie expected to be associated with decisionmaking competence. The ability to acquire knowledge and reason about cognitively complex information, regardless of the domain, should support the development of decision-making skills. Empirically, this conjecture seems to hold true. Work by Bruine de Bruin et al. (2007) showed that scores on RPM were correlated with overall DMC as well as each subcomponent of DMC. This is also corroborated by Del Missier et al. (2012) and Román et al. (2019) who showed that fluid intelligence was linked to several aspects of decision-making competence. In fact, some researchers have argued that decisionmaking competence as a construct is so intertwined with general intelligence that measures of individual differences in decision-making competence are in fact just tapping into general intelligence (e.g., Blacksmith et al., 2019). However, Bruine de Bruin et al. (2020) emphasize that DMC is correlated with decision outcomes even when controlling for fluid intelligence. Nevertheless, it is clear that intelligence is an important predictor of decision-making competence, whose role should be illuminated by incorporating additional measures of cognitive ability in the models.
Although intelligence is arguably important in various decisionmaking domains, some researchers have highlighted that other general cognitive abilities are equally important. Del Missier et al. (2012) found that, in addition to intelligence, measures of EFs contributed to decision-making competence. EF is an umbrella term for different cognitive control processes that are involved in the regulation of cognition and behavior (Miyake et al., 2000;Miyake & Friedman, 2012).
EFs have been studied extensively within cognitive psychology and cognitive neuroscience, and EF has been linked to a multitude of positive outcomes, such as financial outcomes (Ballinger et al., 2011) and mental and physical health (Moffitt et al., 2011). The predominant model of EF was formulated by Miyake et al. (2000) and consists of three interrelated components-inhibition, shifting, and WM/updating. These three components interact during cognitive processing and are subserved by a central executive network in the brain, comprising nodes in the dorsolateral prefrontal cortex and the posterior parietal cortex (Menon & Uddin, 2010;Sherman et al., 2014). These areas of the brain, and the tasks tapping into EF, are often implicated in various neural conditions, such as attention deficit hyperactivity disorder (ADHD) (Diamond, 2005) and Alzheimer's disease (Baudic et al., 2006).
With respect to decision-making competence, studies have found that different components of EF predict different aspects of decisionmaking competence (Del Missier et al., 2010;Del Missier et al., 2012).
For example, inhibition was more tied to the aspect "applying decision rules," whereas shifting was linked to the "consistency in risk perception" aspect (Del Missier et al., 2012). In related work, Rosi et al. (2019) found that WM capacity mediated the age-related decline in "applying decision rules." Thus, all three aspects of EF contribute to decision-making competence.
1.2 | Decision-making competence, spatial, and temporal processing Visuospatial cognitive ability refers to the capacity to perceive and reason about visuospatial properties in the environment, which is another candidate ability that is likely to contribute to better decision making. This visuospatial ability is called upon in multiple levels of spatial processing; playing chess and imagining prospective moves is at a lower level, whereas navigating in an unfamiliar city constitutes a higher level. These levels are interrelated both cognitively and neurologically in the human brain and are grounded in our overall ability to process space (Hegarty et al., 2006). Research has consistently shown that individual spatial abilities vary substantially in the population. In addition, superior visuospatial ability is strongly linked to various positive outcomes, such as achieving professional success in science, technology, engineering, and mathematics (STEM) fields, such as engineering and mathematics that rely heavily on reasoning about visuospatial elements (Wai et al., 2009) and learning mathematics in childhood (Verdine et al., 2014). Research on the impact of visuospatial abilities on decision making is scarce, but the visuospatial ability has been associated with better decision making in the elderly (Gamble et al., 2015;Han et al., 2015). Thus, visuospatial abilities are likely involved in decision-making competence, but the specific relationship is unclear, especially when taking additional abilities into account.
Good decision making is not only about making the right decision among a set of alternatives but also doing it at the right time. In addition, the time it takes until a beneficial outcome is received can be viewed as a cost and is weighted against the value of the outcome (Wittmann & Paulus, 2008). There are individual differences in how people reason about temporal aspects; some individuals are more patient in their choices of saving versus consumption, whereas other individuals seek to ensure immediate gratification. As such, time T A B L E 1 Overview of prior research investigating decision-making competence and their relation to cognitive abilities  management is a pertinent aspect of decision making where the temporal discounting of various choices has to be taken into account (Frederick et al., 2002). Thus, value functions about health and wealth have to factor in when one can expect to reap the benefits of the choices one makes-receiving a dollar today is worth more than receiving a dollar tomorrow. This is an instance of intertemporal choice (Frederick et al., 2002), and studies have shown that there are individual differences in how people estimate the cost of time, where impulsive traits have been linked to a preference toward temporal discounting (Paasche et al., 2019). Related research on consumer behavior has shown that how individuals map objective future time onto a nonlinear subjective perception of time drive intertemporal preferences (Zauberman et al., 2009) and that these preferences can vary depending on whether the consumption concerns material or experiential goods (Goodman et al., 2019). In addition, studies have shown that individuals with higher intelligence (Burks et al., 2009) and higher EF (Basile & Toplak, 2015) were more patient and tended to discount intertemporal choices to a lesser degree. Although studies have established a relationship between temporal discounting and decision making, to our knowledge, no one has investigated whether time perception ability influences decision-making competence.
Time perception has been studied extensively in psychology and cognitive neuroscience, where one pioneering model was named the "Pacemaker-Accumulator-Model" (e.g., Church, 1984;Meck, 1983).
According to this model, the processing of temporal information is carried out in three synchronized systems: One system-"the clock system"-is a dopaminergic system that feeds units, or temporal pulses or neural ticks, through a "gate system" and into an accumulator. The stimulus-dependent information stored in the accumulator is then fed into WM, in prefrontal and parietal cortices, for decision making. Indeed, executive functioning is intrinsically connected with time perception as EFs are involved in planning for the future. Mioni et al. (2017) found a moderate correlation between time perception and WM in children. Some have argued that memory deficits that are manifest in children with ADHD may be caused by an abnormally fast internal counting process that affects time perception ability (Marx et al., 2017). Neurologically, Wittmann and van Wassenhove (2009) have shown that the neurological underpinnings of time perception are very close to those associated with ADHD. Studies have also suggested that everyday multitasking, and by extension successfully managing complex decisions, requires sequential planning and execution of actions that could be supported by spatiotemporal processes (Mäntylä et al., 2017;Todorov et al., 2018). Therefore, we expect that time perception ability is associated with decision-making competence above and beyond the influence of EF, even if they are interrelated constructs.

| Decision-making competence and numerical ability
It stands to reason that being numerate is imperative in today's society in order to make important and informed decisions, such as about one's financial situation and medical issues. These decisions often involve making decisions involving numbers. For example, these numbers may express probabilities, quantities, and frequencies that must be accurately decoded and put into a specific context for the decision maker. Thus, not only must one be able to read and write adequately, but one must also be proficient in understanding and using numbers.
The degree one can do this is called numeracy (Peters et al., 2006;Reyna et al., 2009), and mounting evidence highlights the importance of this skill to decision making. For example, within the domain of health numeracy, findings by Schwartz et al. (1997) demonstrated that greater numeracy was associated with a better understanding of the benefits of mammography. In a similar vein, Peters et al. (2007) found that individuals with lower numeracy struggled more than more numerate individuals in comprehending medical information that was unordered or contained superfluous information. Thus, being more numerate, in terms of understanding percentages and ratios, is linked to understanding the risks and benefits of medical information. Also, numeracy accounts for a unique variance of normative decision making when controlling for effects of education and income (Cavanaugh et al., 2008) and general intelligence (Peters et al., 2006). Numeracy can also serve as a buffer against motivated reasoning when it comes to interpreting political facts (Lind et al., 2018). Work by Woloshin et al. (1999) suggests that individuals with low numeracy may show difficulties using frequencies and percentages in a consistent manner (e.g., transforming information from one format to another) and may encourage these individuals to rely on affective interpretations of information about risks and benefits.
Besides the health domain, research on numeracy and decision making has shown consistent associations between numeracy and decision making involving numbers (e.g., Peters et al., 2006). More numerate individuals are also less susceptible to framing effects and draw more accurate affective meaning from numbers that guide decisions. In terms of the framing effect, it seems that highly numerate individuals can transform numbers from one mode of presentation to another (e.g., percentages to frequencies) more readily and thus have both formats available for comparison. A recent study also found that numeracy was a robust predictor of superior decision making and financial outcomes (Sinayev & Peters, 2015). Thus, extensive research shows consistent links between superior decision making and numeracy that is independent of general intelligence and education. However, it is still unclear how decision-making competence relates to other numerical competencies, such as arithmetic abilities or the innate intuitive sense of nonsymbolic number-the number sense (Dehaene, 2011;Halberda et al., 2008). The number sense refers to the empirical fact that human beings are endowed with an innate domain-specific ability to represent and manipulate quantities (Dehaene, 2011;Gelman & Butterworth, 2005), which is an ability phylogenetically shared with other primates and animals (Dehaene, 2011;Feigenson et al., 2004). It has been suggested that this representational system provides the foundation onto which the culturally acquired symbolic system-mathematics-is mapped (Feigenson et al., 2004). Within the domain of mathematical cognition, the number sense has been extensively studied in terms of how it relates to mathematical abilities (e.g., Halberda et al., 2008;Libertus et al., 2011) and to developmental dyscalculia (e.g., Mazzocco et al., 2011;Rousselle & Noël, 2007;Skagerlund & Träff, 2014). The ability to discriminate between sets of objects has been linked to better mathematical abilities and also to developmental dyscalculia.
Previous studies have found a relationship between normative decision making and affinity with numerical symbols and mapping them to their underlying magnitude (cf. Peters et al., 2008;Peters & Bjälkebring, 2015;Petrova et al., 2014;Schley & Peters, 2014;Sobkow et al., 2020; see also Patalano et al., 2015). We add to the current literature by investigating if acuity of the innate number sense, which is devoid of symbols, is related to DMC. Thus, to get a nuanced view of how overall affinity with numbers influences decision-making competence, we include a set of measures tapping into different aspects of numerical abilities.

| The present study
Our objective was to gain a better understanding of the cognitive underpinnings supporting decision-making competence in adults using a comprehensive test battery that extends previous efforts and insights. A novel contribution resides in including several unexplored predictors that can yield additional knowledge, beyond previously known predictors, about the cognitive abilities subserving decisionmaking competence.
We assessed the relative contributions of general cognitive abilities, such as general intelligence, EF, visuospatial ability, and time perception on different aspects of decision making as well as an overall composite score of decision-making competence. Above and beyond those general abilities, we assessed the contribution of different aspects of numeracy; these aspects consisted of written arithmetic, numeracy (i.e., Berlin Numeracy Test [BNT]), and number sense. To investigate these putative associations, we used hierarchical multiple regression modeling to investigate the relative importance of these cognitive abilities. This allows us to insert control variables, such as age and gender, and incrementally include blocks with previously known predictors, such as intelligence and EF. Although the current approach is mainly explorative, we expect that intelligence, EF, and numeracy to be predictive of decision-making competence, while we remain agnostic as to the role of the other cognitive abilities.

| Participants
The participants consisted of university students enrolled at Linköping University in Sweden and were recruited from varying academic disci- plines. An a priori power analysis, based on a Cohen's f 2 > .15 (a medium-sized effect ;Cohen, 1988), with 11 predictors and an alpha level of .05, and a desired statistical power of 95%, the minimum sample size was calculated to N = 179. Two participants did not complete both sessions (one male and one female), so the final sample (N = 182) consisted of 93 women and 89 men with a mean age of 24.00 (SD = 3.35). All participants had Swedish as their native language and had a normal or corrected-to-normal vision. The participants gave their informed and written consent. The study was approved by the local ethics committee and conformed to the Helsinki convention.
After completing the study, the participants were given $100 as compensation for their participation.

| Assessment of decision-making competence
Decision-making competence was measured using the A-DMC

| Assessment of general cognitive abilities 2.3.1 | General intelligence
To assess general intelligence, we administered a short version of Raven's Advanced Progressive Matrices (RAPM; Raven et al., 1988) developed and normed by Bors and Stokes (1998). This version con-

| EFs
Inhibition was measured using a traditional color Stroop task, which we have used previously (e.g., Skagerlund & Träff, 2014;Strömbäck et al., 2020). Each subject was presented with a piece of paper with 30 color words written in two separate columns. This test was divided into two conditions: one with congruent color words and one with incongruent color words. Each condition was completed twice, resulting in 60 words in each of the two conditions. The subject was to identify and read aloud as fast as possible the color with which the words were written. Immediately upon completion of one round of 30 words of a given condition, the subject went on to the next round

| Visuospatial ability
A mental rotation task was used to assess visuospatial ability, developed by Neuburger et al. (2011). This paper-and-pencil test contained 16 items, consisting of block figures in various configurations, based on earlier work by Shepard and Metzler (1971) and Vandenberg and Kuse (1978). Each item consists of a reference figure, where the reference was located on the left side accompanied by four comparison stimuli located on the right side adjacent to the target. The comparison stimuli always consisted of two "correct" and two "incorrect" items. The primary task was to identify the two matching items and respond by marking them with a pencil. The incorrect items were visually mirrored instances of the correct target.
All comparison stimuli were rotated in the picture-plane and in one of six rotation angles: 45 , 90 , 135 , 225 , 270 , or

| Time perception
The perception of time was measured using a time discrimination task (see Skagerlund & Träff, 2014). The task was to estimate which of two visually presented stimuli was presented the longest. The participant was presented with a reference stimulus (a red ball) centered on the screen, on a white background. The reference stimulus presentation lasted for 3000 ms, followed by a blank screen for 500 ms, after which a target stimulus (a blue ball) appeared centered on the screen. After the target stimulus disappeared, a response screen followed prompting a response. The reference was always presented for 3000 ms and before the target, whereas the target stimuli duration ranged from 1500 to 6000 ms. The participant pressed either the "a" key (marked with red) or the blue-marked "*" key. The test included four practice trials and the participants were asked not to use any counting strategies, such as subvocal counting, during the task. The internal consistency coefficient was calculated to .71.

| Arithmetic ability
Arithmetic calculation ability was assessed using a similar procedure as Gebuis and van der Smagt (2011) and Lindskog et al. (2017). This test was divided into four subtests (addition, subtraction, multiplication, and division). For each subtest, the participants were presented with a sheet of paper containing printed arithmetic problems of increasing difficulty. For each subtest, they were instructed to complete as many problems as they could within the allotted time of 120 s. A brief pause was included in between each subtest. The difficulty level of the problems was manipulated by increasing the number of digits or by requiring borrowing or carrying. Each subtest contained 54 problems, except for the division task that contained 27 problems. In sum, the maximum score was 189. The total number of correctly solved arithmetic problems across all four conditions was used as a measure of arithmetic calculation ability. Cronbach's alpha was calculated by checking the internal consistency across each arithmetic subtest and resulted in α = .88.

| Number sense
Two arrays of blue and yellow dots, ranging from 5 to 21, were displayed on a computer screen. The participant was then asked to determine, as quickly and as accurately as possible, which array contained more dots and press a corresponding color-coded key.

| Procedure
The study was conducted over three sessions in order to avoid fatigue. The first was a computerized survey session containing the administration of the A-DMC, which was filled out at the same time as other participants in a group setting. Then two individual one-onone sessions with an experimenter at a research lab on the university campus, each lasting about 45 min. All testing was completed within one month during the spring of 2017. Instructions were written in the survey session and read aloud from a printed manuscript during the one-on-one lab sessions. All tests were administered in the same order for all study participants. Computer-based tasks administered during the one-on-one session were run on a laptop, using SuperLab PRO 5. During the first individual lab session, numeracy measures were administered alongside additional questionnaires and tests not relevant for the current paper (e.g., questionnaires regarding personal finance and donating behavior). During the second individual session, the following tasks were administered: RPM, visuospatial task, time perception task, and the EF tasks (nonrelevant tasks regarding emotional processing were also administered during this session).

| Analytic approach
Prior to data analysis of computerized tasks, intraindividual trials were examined to remove outliers; RTs < 200 ms were removed, as were RTs > 2.5 SD of the individual within a test. For the multiple regression models we verified that, for each model, multicollinearity was not an issue. The variance inflation factor (VIF) was always below 2.11, and tolerance was above .48 for all models. Table 2
The complete model explained a total of 39% of the variance, R 2 = .39, F(11, 170) = 9.78, and p < .001. Analysis of individual predictors of overall decision-making competence is summarized in Table 3.
By computing squared part correlations (i.e., semipartial correlations), we can determine the unique variance explained by each individual predictor. Looking at the final model in Step 4, we found that age did not explain any unique variance, whereas gender accounted for 4% unique variance, favoring males. Within the block containing general cognitive abilities, general intelligence contributed 6% unique variance, which was also the strongest predictor of overall decision-

| Resistance to framing
Recognizing the amount of space required to present the stepwise inclusion of all individual blocks in all the regression analyses, we only present the final models for each subcomponent in Tables 4 and 5 for brevity. The multiple regression analysis of resistance to framing can be found in

| Resistance to sunk costs
The multiple regression analysis of resistance to sunk costs can be found in Table 4. Age and gender accounted for 7% of the variation in

| Consistency in risk perception
The multiple regression analysis of consistency in risk perception can be found in Table 4. Age and gender accounted for 10% of the variation in consistency in risk perception, R 2 = .10, F(2, 179) = 9.83, and p < .001. Adding a block of general cognitive abilities added an additional 7%, ΔR 2 = .07, F change (5, 174) = 2.97, and p = .013. Adding the numerical block did not add to the model, ΔR 2 = .02, F change

| Underconfidence/overconfidence
The results from the regression analysis of under/overconfidence aspect of A-DMC can be found in Table 5. The block containing age and gender did not explain any significant variance, R 2 = .03, F Overview of individual predictors of under/overconfidence is summarized in Table 5. No cognitive predictors explained any unique variance except for visuospatial ability, which was negatively associated with under/overconfidence (6%). Time perception was marginally significant as a predictor (p = .066) and contributed with 2% unique variance explained.

| Recognizing social norms
The regression models of recognizing social norms can be found in Table 5. Age and gender did not explain any significant variance,

| DISCUSSION
Our objective was to investigate which cognitive abilities that support decision-making competence in adults. By using a comprehensive test battery of cognitive abilities, our hierarchical multiple regression analyses enabled us to carefully control for previously known predictors and control variables while evaluating the impact of hitherto unexplored, but potentially important, variables of interest.
Regarding the overall decision-making competence, consisting of a composite score of the individual components of the A-DMC, the strongest predictor was general intelligence (RAPM). This was in line with our initial expectation and corroborates earlier studies by Bruine de Bruin et al. (2007) and Del Missier et al. (2012). Thus, we reinforce the notion that successful decision making is bolstered by having the ability to reason about cognitively complex information, irrespective of domain.
What was perhaps more interesting, however, was to examine which, if any, other cognitive abilities would predict overall decisionmaking competence once including general intelligence in our models.
Earlier research by Del Missier et al. (2012) found that EF could explain additional variance above and beyond general intelligence.
However, we failed to find a significant contribution of any EF component to overall decision-making competence. Although the WM component was a significant predictor in Step 2 of the analysis (see Table 3), this contribution disappeared once we included the numerical block of predictors. Thus, it seems that the combination of numerical abilities and visuospatial abilities outcompeted WM ability in the final model, which also highlights the importance of utilizing comprehensive test batteries such as the one in the current study. Still, WM was the strongest predictor of "application of decision-rules" component as predicted, which is in line with Del Missier et al. (2012). WM likely ties into this aspect of A-DMC insofar as having a greater WM capacity will allow an individual to keep more information active in a mental workspace, thus supporting effective application of decision rules.
We also explored whether the order in which we inserted the predictors could be a factor. Therefore, explored whether putting the EF measures first before any other cognitive variables would change our models. However, this had no impact on the results (see Tables S6-S8). Considering the fact that one can use different measures of the trailmaking task and Stroop task, we also performed exploratory post hoc analyses of derived difference and ratio scores of the trailmaking test and Stroop task as opposed to the absolute response times (see Arbuthnott & Frank, 2000;Salthouse, 2011;Sánchez-Cubillo et al., 2009, for discussions on which indices to use).
Using difference scores (e.g., B-A in the trailmaking task and congruent color condition time subtracted from the incongruent condition) revealed no change in results (see Tables S1-S3 Tables S1-S5). These post hoc analyses should be interpreted with caution given that our a priori decision was to use the absolute scores on the trailmaking task and the Stroop task in order to account for the speed and efficiency with which one can switch between aspects and inhibit irrelevant responses. Taken together, the overall pattern of our data suggests that EF are still important factors of DMC, although the extent to which should be further investigated, but that other predictors could be potentially be even more important.
Considering the role of visuospatial cognitive processing, as indexed using a mental rotation task, it is likely that visual and spatial aspects of WM are critical elements in our observed association with decision-making competence. By examining the numerical block of predictors that targeted slightly different aspects of mathematical competency, we observe that being able to rapidly solve arithmetical problems did not predict decision-making competence. Although raw calculation efficiency using numerical information proved not to be important, the ability to reason and solve more complex problems containing numerical information was (i.e., numeracy). This corroborates studies on normative decision making (Cavanaugh et al., 2008;Peters et al., 2006;Sinayev & Peters, 2015). One caveat, however, is that the arithmetic calculation task and the numeracy task involve different levels of time pressure in terms of time required per item. This could result in different levels of perceived stress on these task that could impact performance unevenly.
Therefore, our study reinforces the preexisting notion that numeracy is indeed important for decision-making competence. Individuals with higher numeracy likely decode numerical information faster, but more importantly, these individuals can apply correct mathematical concepts and procedures and make judgments about which parts of numerical information are pertinent to the situation at hand.
The link between numeracy and normative decision making involving numerical information is uncontroversial, but some argue that numeracy is tied to nonnumerical decision making as well (Cokely et al., 2018). Numeracy has been suggested to relate to judgment of social relationships, behavioral norms, health behaviors, among others (Cokely et al., 2018). Although numeracy was primarily related to "application of decision rules," which involves numbers, our current results support the idea that numeracy assists decision making beyond the numerical domain. Interestingly, numeracy predicted decisionmaking competence even when general intelligence was included in the model, which demonstrates a demarcation between these two constructs. Considering the mounting evidence showing a strong correlation, depending on the population and measures used, between general intelligence and numeracy (r's mainly ranging from .37-.61; e.g., Cokely et al., 2012;Bruine de Bruin et al., 2007;Dieckmann et al., 2015;Sobkow et al., 2020) our current results reinforce the notion that these abilities and skills contribute to decision making independently of each other. Still, when considering the unique contributions of each predictor, as indexed by the squared part correlations, general intelligence is a stronger predictor than numeracy. The prominent role of general intelligence vis-à-vis numeracy is also reflected by the fact that intelligence was related to four out six individual components of the A-DMC, which is in stark contrast to the numeracy measure that was related primarily to the "application of decision rules." To further investigate the roles of numeracy and general intelligence, we performed exploratory post hoc regression analyses in order to disentangle the relative role of each predictor. Given that we inserted a block of general cognitive abilities prior to inserting a block containing numerical predictors, we were aware of the fact that general intelligence may have explained common and unique variance that the numeracy question would explain as well. Therefore, we reversed the order of the blocks, such that the numerical block was inserted first and the general cognitive block inserted afterward. Nevertheless, the results remained identical to our initial models, where numeracy showed a pr 2 = .02 (ß = .18) and general intelligence showed pr 2 = .06 (ß = .31). Thus, the results indicate that general intelligence shows a stronger link to overall decisionmaking competence, both in terms of beta values and squared part correlations, than numeracy, which goes against Cokely et al. (2018). We further investigated the relationship between numeracy, intelligence, and decision-making competence through a series of mediation analyses to fully see whether one predictor was fully mediated by another. However, we found both significant direct and indirect effects of numeracy and intelligence on decision-making competence, regardless of whether the predictors were assigned as the independent or mediating variable. For brevity, we refer to supporting information (see Figure S1) for further details on the mediation analyses. We therefore conclude that both numeracy and intelligence have an independent impact on decisionmaking competence.
The last component of the numerical skills that we included in our models was the measure of the innate number sense (Dehaene, 2011;Halberda et al., 2008). The number sense allows humans to estimate quantities and constitute a system for semantic representation of numerosity. The number sense is responsible for representing large and approximate numbers via a logarithmic analog mental number line that is inherently noisy (Dehaene, 2011;Feigenson et al., 2004).
Therefore, the ability to discriminate between sets of objects is ratio dependent (Halberda et al., 2008) and would most likely be a subcomponent in the construct of numeracy given that numeracy itself concerned with the ability to reason about probabilities and fractions (Cokely et al., 2012). But the direct relationship to decisionmaking skills has remained unexplored. Our results indicate that number sense is not related to any specific aspect of the A-DMC.
Interestingly, however, it was significantly predictive of overall decision-making competence (ß = −.15, pr 2 = .02), which is very similar to the effect of numeracy on A-DMC. The mechanism by which number sense operates on decision making is unknown, but a tentative interpretation is that having a more refined number sense allows one to make accurate, intuitive estimations of quantities and probabilities in many areas of everyday life. The ability to discriminate between sets of objects, the number sense, is ratio dependent and obey psychophysical laws, such as Weber's law (Halberda et al., 2008). As such, it is conceptually plausible that the number sense constitutes a component of the construct of numeracy, given that numeracy itself is concerned with the ability to reason about probabilities and fractions (Cokely et al., 2012). Nevertheless, it is interesting that the number sense predicts decision-making competence concurrently with numeracy in our models.
The final ability that we investigated was time perception ability. Time management is a pertinent aspect of decision making where the temporal discounting of various choices has to be taken into account (Frederick et al., 2002), but no previous study has investigated whether the specific ability to perceive time is related to sound decision making. It stands to reason that an accurate representation of time would be beneficial for navigating in a decisionmaking landscape, where both accurate and timely decisions have to be made. We inserted this predictor last in our models in order to be as conservative as possible and to investigate the specificity with which it could predict decision-making competence. Despite this conservative approach, we found that time perception ability predicted overall decision-making competence and was related to two specific aspects of the A-DMC components. In fact, time perception ability was a stronger predictor than both number sense and numeracy (ß = .18, pr 2 = .03), thus being outperformed only by general intelligence out of all predictors. Moreover, time perception as a construct has been linked to other cognitive abilities, such as WM capacity (Mioni et al., 2017). Supra-second intervals, especially at the higher end of the temporal spectrum, would place a significant load on WM resources. Indeed, supra-second temporal processing activates prefrontal areas of the brain (Wittmann, 2009), which in turn are cortical loci for spatial WM processing together with connecting areas in parietal regions (Rotzer et al., 2009;Rubinsten & Henik, 2009). It is therefore notable that time perception ability was a significant predictor, despite including WM capacity and other EFs in the models. How can time perception ability bolster decision making? A plausible mechanism is that time perception ability relates to temporal discounting and self-regulation, such that individuals with well-developed time perception assign value functions to decisions that favor long-term intertemporal choice. Also, work by Todorov et al. has suggested the importance of spatial and temporal processing in successful everyday multitasking performance (Todorov et al., 2018) and that the coordination of several tasks requires time-to-space transformational processes (Mäntylä et al., 2017). The coordination of everyday multitasking was suggested to be reliant upon temporal coordination of activities, where the representation and planning of these activities can spatially represent to enhance performance (Mäntylä et al., 2017). This interpretation could unify our findings where both temporal processing and spatial processing are linked to DMC. Therefore, a novel contribution of this study is establishing that temporal and spatial processing is also tied to scores on the A-DMC. The link between ability and intertemporal choice remains to be empirically investigated, however.
Limitations of the current research include the restricted way of measuring cognitive abilities. For any specific cognitive ability, there are several viable tools that can be employed in order to assess these abilities. For instance, we used three instruments to measure the three respective EF, and although these are validated and conventional assessment tools, one could have substituted them in favor of other measures of EF. One could utilize another WM task specifically targeting the updating component of WM and use more than one indicator to assess each facet of EF to overcome task impurity and other measurement problems. Therefore, we cannot make definitive claims about EF and the relationship with DMC, given the observed absence of an effect, but our results nevertheless provide support for the importance of auxiliary processes beyond EF. It is also worth noting that all predictors in our models are quite modest in terms of their explained variance, joint as well as unique. This also ties into another limitation, although by design, which is the sole focus on cognitive abilities. Decision-making competence is a multifaceted construct that is related to, among others, personality variables (e.g., introversion) and decision-making styles (e.g., brooding about decisions). In a recent paper by Bruine de Bruin et al. (2020), the authors suggested that emotion regulation and motivation are important psychological processes that complement general cognitive abilities in making sound decisions. Moreover, the authors underscore that certain decision-making components may be subject to change throughout ontogeny, such that some aspects become weaker as a function of age (e.g., fluid intelligence Another limitation of our study is that the study sample consisted of university students who likely are homogenous in terms of their socioeconomic status and overall cognitive abilities and decision making. Additional research using more representative samples across a wider age range could provide data with more variance, and thus perhaps a more comprehensive picture of which cognitive abilities are important for decision making. Utilizing a larger sample could also prove beneficial in detecting small effects and contributions of specific cognitive abilities. Taken together, our main findings from administering a comprehensive test battery of cognitive abilities are threefold. First, we reinforce the important role that general intelligence plays in decision-making competence and that numeracy and intelligence show independent contributions. Second, EFs, which have been tied to A-DMC scores in previous research (e.g., Del Missier et al., 2012), did not contribute to superior decision-making competence once numerical, temporal, and spatial abilities were included in the models. Third, a novel finding is that time perception was tied to superior decision-making competence, which invites further research into exactly how this ability relates to decision-making competence.
Overall, this study extends previous work by Del Missier et al. (2012) and related research probing which cognitive abilities matter for successful decision making.
understand everyday decision making and its underlying processes. A primary focus is to understand decision making and priority setting in health care.
Daniel Västfjäll is a professor of psychology at Linköping University. His main interest is in emotion and its effect on everyday behaviors. He is a co-director of the Judgment, Emotion, Decision, and Intuition Lab (JEDI Lab) and affiliated with Decision Research in Eugene, Oregon, USA.

SUPPORTING INFORMATION
Additional supporting information may be found online in the Supporting Information section at the end of this article.