Introduction

Cognitive differences between males and females are a broadly researched topic. Over the course of the years, research on gender differences in cognitive abilities has received criticism about reproducing and enhancing social stereotypes. For instance, it has been reported that males tend to outperform females on typical spatial tasks like mental rotation (e.g. Linn & Petersen, 1985; Richardson, 1994; Masters & Sanders, 1993; Terlecki et al., 2008), whereas females show relatively better performance on tasks like verbal fluency in some studies (e.g. Burton et al., 2005; Weiss, Ragland, Brensinger et al., 2006). Especially within the domain of spatial cognition, the impact of gender has been included in many studies.

Although such differences can be present to some extent, sources of bias such as participant recruitment, sample representation, task selections, as well as testing conditions have had a significant impact on the direction of the findings (Caplan et al., 1997). Currently, we are aware that these gender differences are more specific and refined, mediated to a large extent by factors such as education, experience, environment and culture (Jäncke, 2018; Hyde, 2016).

In line with the above, stereotypes concerning gender and spatial functioning appear to be prevalent, with a general tendency to assume a male advantage in spatial tasks (see e.g. Hyde, 2016; van der Heyden et al., 2016). Apart from the question of whether such stereotypes are in accordance with documented gender differences, they have been found to pose a negative impact on cognitive functioning and perception of one’s own performance (for review see Ellemers, 2018). This effect is termed a stereotype threat, which means that the group that is subject to a negative stereotype, e.g. ‘women are worse in math’, may show lower performance, in accordance with the stereotype (e.g. Spencer et al., 1999; McGlone & Aronson, 2006, Campbell & Collaer, 2009). Its counterpart, the stereotype boost, has also been reported, where those benefiting from the stereotype improve in performance (e.g. Armenta, 2010; Shih et al., 2012). To gain insight into how extensive such effects may be, a more detailed examination of stereotypes in spatial cognition is in order. This would provide a valuable addition to understand gender differences in spatial cognition and the impact of related social factors. The aim of this study is therefore to further clarify which stereotypes exist for spatial abilities and how individual differences concerning age and gender affect the direction and strength of such stereotypes. We will examine stereotypes concerning spatial cognition, in combination with how individuals perceive their own performance, relative to their peers. Additionally, we will study how such stereotypes and self-perception relate to gender and age of the participants.

Objectively, the impact of gender on spatial cognition is more fine-grained than simply stating that a male advantage is present. An important distinction has been made with regard to studying individual differences in spatial cognition; between spatial abilities in a small and large scale (Hegarty et al., 2006). Small-scale spatial abilities consist of tabletop paper and pencil tests and therefore reflect the spatial domain within one’s reach. In contrast, orientation of oneself and navigation occur in large scale space. As shown by Hegarty et al. (2006) small scale spatial ability and larger scale navigation and orientation are at least partially dissociated. For gender, perhaps the most convincing difference between males and females is found for mental rotation (e.g. Masters & Sanders, 1993). However, it should be noted that spatial experience has been shown to positively impact mental rotation performance (e.g. Terlecki et al., 2008) and may well show a gender bias, favoring males. Furthermore, a male advantage is not evident for all forms of small scale spatial ability. Depending on object identity, spatial location memory shows a female advantage (e.g. Voyer et al., 2007; Tottenham et al., 2003). Strategic differences are found for the large scale spatial skill of navigation, where males tend to outperform females when metric information is needed, but comparable performance is found when landmarks are available as well (Grön et al., 2000). Taken together, an overall statement about gender effects in spatial cognition is difficult to make. For both small scale ability and navigation male and female advantages have been found, as well as absence of such gender differences, depending on the specific task content at hand.

In addition, age has shown to be a major factor in spatial performance. Many spatial functions decrease with increasing age. Exemplary for this fact is the recent surge of interest in navigation measures as a potential early predictor for pathological aging. A number of recent cross-sectional studies highlight the impact of aging from a relatively young age onwards (e.g. Coutrot et al., 2018; Van der Ham et al., 2020). Yet, stereotypes for spatial cognition appear to focus mainly on gender, instead of age, and very little documentation is available concerning potential age related stereotypes. Therefore, in this study we also examine potential age related stereotypes, as the objective evidence for the impact of age is much stronger, compared to the impact of gender.

In recent work (van der Ham et al., 2020), we observed a substantial impact of both age and gender on self-reported performance on navigation. Males and older individuals tended to overestimate their own performance, whereas females and younger individuals showed underestimation. As stereotypes have been shown to impact self-reported performance (Ellemers, 2018), these two measures may be related. To create a comprehensive view on stereotypes in spatial cognition, we include measures of stereotypes both for gender and age (males and females, younger and older individuals), and self-reports in relation to peers in terms of gender or age. It could be that such effects of age and gender may also be present for the extent stereotypes are held by individuals.

Thus, here we study the prevalence of stereotypes for both small scale spatial ability and navigation skills. Given the traditional focus of research on small scale spatial ability and gender, a stronger gender stereotype favoring males is expected for this domain, compared to navigation. With regard to gender and age, both stereotype threat and stereotype boost may be at play. If stereotype threat is the dominant effect, stereotypes may be stronger in females, due to their lower self-reports. If stereotype boost prevails, males may hold the stronger stereotype. For age, stereotype threat would entail a stronger stereotype for older individuals, whereas stereotype boost would show a stronger stereotype for younger individuals. Furthermore, in line with earlier findings, we expect higher self-reports for males and older individuals, in comparison to females and younger individuals, respectively.

A large sample exploration of the occurrence and extent of stereotypes was performed. With a questionnaire in a heterogeneous sample of Dutch participants we explore both stereotypes and self-reports on small scale spatial abilities and navigation ability, in relation to gender and age. As the gender (male, female) and age categories (young, old) may be perceived differently by individuals, the extent to which individuals identified with these gender and age groups was also considered. In other words, we included to what extent the participants felt male or female and young or old.

Methods

Participants

In total 985 Dutch participants filled out the questionnaire, with an age range of 18–94 years. The sample consisted of 467 males, 513 females and 5 nonbinary participants. Due to the gender-related questions and the large difference in group size, the nonbinary participants were not considered in the analyses, resulting in a final sample of 980 participants. The demographic characteristics of the sample, split up by gender and age group, are provided in Table 1. A one-way ANOVA show a significant effect of education level between the different age groups, F(3,976) = 17.69, p < .001. Higher education levels were found for the 18–30 and 31–45 year olds, in comparison to both 46–60 and > 60 year olds (p < .05 in all cases). The study was approved by the local ethical committee at Leiden University, and in accordance with the declaration of Helsinki (2013) each participant provided informed consent on the first page of the online questionnaire portal, prior to starting the questionnaire itself. Individuals with neurological or psychiatric conditions were explicitly asked to abstain from participation. Recruitment occurred through various channels; e.g. social media and personal networks.

Table 1 Descriptive statistics of all participant subgroups, split up by gender and age group
Full size table

Materials and procedure

Participants first read and digitally signed the informed consent, after which they filled out a questionnaire presented through Qualtrics XP software. The questionnaire consisted of demographic questions, a set of questions related to stereotypes, and a set of questions concerning gender and age. All questions were phrased in Dutch and aimed at Dutch participants.

Participants were asked about their gender, age, education level (1–7 low to high, Verhage, 1964), and other demographics not included in the current analyses. For the stereotype questions, a distinction was made between spatial ability, as defined by smaller scale, basic spatial processing and navigation ability, large scale spatial processing. Also, a distinction was made between stereotypes of gender and age groups in general, and participants’ ratings of their own performance in relation to their gender and age. Participants were presented with a detailed definition of the term spatial ability prior to the questions related to this term to ensure proper and uniform understanding. The description included the following examples: equally dividing a cake, parking a car, using a triangular ruler, loading a moving van. Next, they were asked how they thought gender affected spatial ability by presenting a ruler with ‘male’ on the left, neutral in the center, and ‘female’ on the right, on which they could indicate if and how strongly a gender effect was present in their opinion. They were asked to place a mark on the ruler in response to the question ‘Who has better spatial abilities?’. For all ruler questions, the response resulted in a score between 0 (most leftward point of the ruler) and 100 (most rightward point of the ruler), which proportionally reflected their rating, with 50 (center) being fully neutral. This question was repeated with ‘young people’ on the left and ‘old people’ on the right. Next, they were asked about how they felt their own performance was in relation to their own gender (a ruler response with left: very bad, right: very good), and then their own age group (a ruler response with left: very bad, right: very good). These four questions were then repeated for navigation ability. A detailed description of this term was also provided, with examples concerning finding the way, identifying one’s location, reading a map, and using a GPS system. In short, this resulted in scores reflecting stereotypes at group level and an estimation of one’s own performance, for small scale spatial ability and large scale navigation ability, with regard to both gender and age.

Lastly, gender and age related questions were asked to operationalize gender identification and age identification. For gender identification, the following question was used: ‘in general I feel’ (slider response with left: male, right: female). For age identification, the same questions were used, with left: young, right: old as response options. Again, all responses were recorded on a 0 (left) – 100 (right) range, with 50 reflecting the neutral center. This completed the questionnaire, after which participants were thanked or their participation and were provided with contact information, in case of questions or comments.

Analyses

First, the occurrence of stereotypes and self-report bias was measured, defined as a score different from the neutral 50 center of the rulers used. For each question, for both gender and age, for both spatial and navigation ability, and for group and self-related estimates, the mean score was compared to 50 with Bonferroni corrected one sample t-tests (alpha was set at 0.05/8 = 0.00625).

Next, the age and gender characteristics of the sample were taken into account in a MANCOVA with age group and gender as between subjects variables, each of the eight ruler questions (4 stereotype, 4 self-report and 4 age-related, 4 gender-related) as the dependent variables, and education level as covariate. Significant interaction effects were followed up with Bonferroni corrected post hoc tests. To also assess whether ratings for each participant subgroup were significantly different from the neutral score of 50, at the middle of the scale a series of one sample t-tests were performed (Bonferroni corrected alpha set at 0.05/64 = 0.00078).

In addition to age group and gender as categorical variables, the questionnaire also included questions concerning the extent to which participants identified themselves based on gender and age. The extent to which participants identified as male or female and young or old was added to the MANCOVA performed for gender and age group. To this end, participants were divided into high and low identifiers, in which high and low refers to the level of agreement with actual gender. A gender identification score was calculated by first calculating the difference between actual gender score (0 for males, 100 for females) and the score provided on the ruler. Next, a split mean procedure was followed. Those with smaller than average difference between their gender identification score and actual gender were considered to identify highly with their own gender, whereas those with a larger than average difference, were marked as low identifiers. For age, it was determined for each participant if they felt older or younger than their actual age. To provide an estimate for this, the ruler response calculation was adjusted to cover the range of 18–100 years of age, where 18 years was 0 (young extreme) and 100 years 100 (old extreme), to cover the full range of adulthood with the scale (age range of participants was 18–94). If the ruler response was lower than the corresponding value for their actual age, participants were considered to provide a young estimate. If it was higher, they were categorized as providing an old estimate. The MANCOVAs for age group and gender, with education level as a covariate were repeated, once for gender identification (high vs. low) and once for age estimate (young vs. old). All additional significant effects are reported (Bonferroni corrected alpha set at 0.05/64 = 0.00078).

Results

Figure 1 reflects the occurrence of stereotypes for each of the questions asked. Bonferroni corrected one sample t-tests were used to assess whether they differed significantly from the neutral score of 50. This was the case in all but one score; p < .001 in all cases except for the effect of age for navigation ability, at group level. The group level measures showed participants favored males for both spatial and navigation ability, and young individuals for spatial ability. Furthermore, all self-reported scores indicate a substantial bias towards better performance in comparison to others of one’s own gender and age. In Table 2 the mean score for each of the participant subgroups (male/female, 4 age groups) and each condition is provided. The outcome of the one sample t tests, comparing each score to the neutral 50 score is integrated in the table with an * for all p’s < 0.0001.

Fig. 1
figure 1

Mean scores for all questions for A ratings at group level and B relative self-reported performance. Error bars represent standard error to the mean

Full size image
Table 2 Mean scores for each participant subgroup, split up by gender and age group, for all questions asked
Full size table

Stereotypes

For the stereotype scores at group level regarding the role of gender, the MANCOVA showed that males were more in support of a male advantage than females as their score was significantly lower for both spatial ability, F(1,971) = 85.58, p < .001, ηp2 = 0.081, and navigation ability, F(1,971) = 77.85, p < .001, ηp2 = 0.074. A main effect for age group was also found here, for both spatial ability, F(3,971) = 4.82, p = .002, ηp2 = 0.015, and navigation ability, F(3,971) = 3.61, p = .024, ηp2 = 0.010. In both cases the 31–45 year-olds scored significantly lower than the > 60 year olds (p < .05). This indicates the stereotype favoring males is stronger for younger compared to older adults. No significant interaction effects were found.

The stereotype scores at group level with regard to age showed a significant effect of gender only for the navigation ability question, F(1,971) = 11.33, p = .001, ηp2 = 0.022, in which males scored higher than females. This indicates a stronger stereotype favoring young individuals for males in comparison to females. A significant effect of age group for spatial ability, F(3,971) = 9.84, p < .001, ηp2 = 0.030, and for navigation ability, F(3,971) = 7.16, p < .001, ηp2 = 0.022, showed that scores were lower for the younger groups in comparison to the older (spatial ability: 18–45 vs. > 45; spatial navigation 18–45 vs. 46–60, all p’s < 0.05). The stereotype favoring young individuals therefore appears stronger for the younger age groups. No significant interaction effects were found.

Self-reports

The self-reported performance levels with regard to gender showed a main effect of gender for both spatial ability, F(1, 971) = 22.21, p < .001, ηp2 = 0.022, and navigation ability, F(1,971) = 43.07, p < .001, ηp2 = 0.042. Males reported higher scores than females in both cases. Age group also showed a significant main effect in both questions, F(3,971) = 5.11, p = .002, ηp2 = 0.016, and, F(3,971) = 9.94, p < .001, ηp2 = 0.022, respectively. In both cases the youngest participants (18–30) had significantly lower scores than the older participants (> 30 years). This indicates that males and older individuals estimate their own ability as higher, in comparison to their own gender. No significant interaction effects were found.

Lastly, the self-reported performance levels concerning age showed a main effect of gender for both spatial ability, F(1, 971) = 39.25, p < .001, ηp2 = 0.039, and navigation ability, F(1,971) = 50.57, p < .001, ηp2 = 0.050. Males reported higher scores than females in both cases. Age group also showed a significant main effect in both questions, F(3,971) = 4.84, p = .002, ηp2 = 0.015, and, F(3,971) = 10.32, p < .001, ηp2 = 0.031, respectively. In both cases the youngest participants (18–30) had significantly lower scores than the older participants (> 30 years). For navigation ability, the interaction effect of age group and gender was significant, F(3,971) = 3.58, p < .05, ηp2 = 0.011, indicating that for females 18–45 year-olds scored lower than > 46 year-olds, whereas for males, only 18–30 year-olds scored lower than older participants. This indicates that also with regard to their own age group, males and older individuals estimate their own ability as higher.

Gender identification

For males, the mean gender identification score was 6.86 (sd = 11.9, range 0-100), which was very close to the male extreme of the scale (0). 4 participants scored below the middle of the scale (50), reflecting a more female than male identification. The mean split performed for the total of 467 male participants resulted in 322 participants in the high gender identification group (score range 0–6) and 145 participants in the low gender identification group (score 7-100). For females the mean gender identification score was 93.9 (sd = 11.9, range 7-100), which is near the female end of the scale (100). Four participants score below the middle of the scale (50), reflecting a more male than female identification. The mean split performed for the total of 513 female participants resulted in 134 participants scoring below the group mean of 94, and 379 scoring above it (Table 3).

Table 3 Mean scores for each participant subgroup, split up by gender and age group, for all high and low subgroups based on gender identification and age identification scores
Full size table

The MANCOVA including age group, gender, and gender identification group (high, low) as between subject factors and education level as covariate lead to the following findings with regard to the scores on the age and gender related questions. Only significant effects including gender identification group are reported here. Overall, gender identification group interacted significantly with gender, F (8,956) = 3.74, p < .001, ηp2 = 0.030. This interaction reached significance for gender stereotype, F(1,963) = 12.96, p < .001, ηp2 = 0.013, and gender-referenced self-rating, F(1,963) = 8.24, p < .005, ηp2 = 0.008 for spatial ability, and for gender-referenced self-rating for navigation, F(1,963) = 9.75, p < .005, ηp2 = 0.010. For age-referenced self-rating for spatial ability (p = .007) and gender stereotype for navigation (p = .008) this interaction effect reached trend level. For all cases the pattern of the interaction was highly comparable; a significant difference in responses between the high and low gender identification group was only found for males. The male low gender identification group performed comparable to both the high and low gender identification groups in females. The male, high gender identification group showed significantly stronger gender stereotypes for spatial ability (p < .001) and navigation (p = .001) and higher gender-referenced self-ratings for spatial ability (p = .002) and navigation (p = .001). Additionally, age-referenced self-rating for spatial ability was also higher in the male, high gender identification group (p = .008).

Age identification

The MANCOVA performed included age group, gender and age estimation group (young, low) as between subject variables, and education level as covariate. All significant effects including age identification group are reported. There was an overall main effect of age estimate at trend level, F(8,950) = 3.43, p = .001, ηp2 = 0.028. Scores were higher for those feeling younger, compared to those feeling older. Specifically, this effect reached significance for spatial ability self-reports, with regard to age, F(1,974) = 12.50, p < .001, ηp2 = 0.013, and to gender, F(1,974), p < .001, ηp2 = 0.014. In both cases, those who feel younger, rate their own performance as higher, when comparing themselves to their own age and gender groups (Table 3).

Discussion

Cognitive stereotypes with regard to gender can have a substantial impact on cognitive functioning through stereotype threat and well as stereotype boost effects (e.g. McGlone & Aronson, 2006; Ellemers, 2018; Shih et al., 2012). Here, we focused specifically on stereotypes with regard to spatial cognition. As age affects navigation performance strongly, stereotypes with regard to age were also examined. The aim of this study was to first gain a detailed picture of the extent and direction of stereotypes held in general population. Next, we examined how individual differences within the population, in terms of age and gender, may affect the strength and direction of the stereotypes held. Additionally, we included self-reported performance in our analyses for two reasons. Previous work has shown a strong impact of both gender and age on those self-reports, and stereotypes could well affect how one perceives their own performance level.

First, we expected a strong stereotype favoring males for spatial ability, more so than for navigation. The results show that a clear male advantage is indicated by the participants, for both spatial ability and navigation. The extent of the stereotype is comparable for both spatial domains. This is not in line with objective findings, as a clear male advantage is mainly found for certain aspects of spatial ability (e.g. Linn & Petersen, 1985), not for spatial navigation (e.g. van der Ham et al., 2020a, b; Grön et al., 2000). Potential age stereotypes have received very little attention in literature, therefore we did not have specific hypotheses concerning this phenomenon. The analyses indicate that such a stereotype exists, but only for spatial ability. Generally, it is assumed that younger individuals have better spatial ability then their older counterparts. For navigation, however, no such effect is present. This effect is in contrast with the fact that especially for navigation ability, a clear negative impact of older age is consistently found (e.g. Moffat, 2009; Gazova et al., 2012; Wiener et al., 2013; Coutrot et al., 2018).

We examined how these stereotypes were affected by gender of the participant. It was clear that males more strongly held stereotypes, in comparison to females. For both spatial ability and navigation, they indicate a stronger male advantage. Moreover, their indication of an age effect favoring younger individuals, is stronger than for females. So, beyond their own group they indicate a stronger stereotype for age. This suggests that a stereotype boost effect may be at play here for the male participants. When examining the self-reported performance scores, this becomes more evident. Notably, all participant groups rate themselves as above average. This effect is particularly strong for the male participants, regardless of whether they compare themselves to their own gender or their own age group. Females thus hold weaker stereotypes, but still show significantly lower self-ratings of performance. These findings are in line with a range of reports on gender effects on self-efficacy concerning general cognitive tasks, where self-efficacy is typically higher in males (e.g. Busch, 1995; Junge & Dretzke, 1995; Huffman et al., 2013).

For age, we see that young individuals hold stronger gender stereotypes, favoring males, in comparison to older individuals. In short, this indicates that those who are favored by the stereotype held, are those who hold the stereotype the strongest, for both males and young individuals. Albeit that the age related stereotypes are weaker. However, when examining the self-reports, we again find stronger overestimation in older individuals, in the opposite direct of the stereotype held. This would argue against a stereotype boost effect for younger individuals, or a stereotype threat for older individuals. In comparison to the gender effects, this could possibly be explained by the limited strength of the stereotype itself.

To further study the impact of gender, we also explored if a nonbinary approach to gender could enlighten the discussion. Therefore, we generated a gender identification score, based on how where a participant placed themselves on the male-female spectrum. First, we found a very strong binary identification pattern. Based on the scores, participants were considered high or low gender identifiers. In addition to the general pattern, this showed that the male pattern of stronger stereotypes and higher self-ratings was even stronger for male, high gender identifiers, in comparison to all others. Therefore, it appears that the strong stereotypes and high self-ratings could possibly be considered as ‘highly masculine’. This conclusion is in line with previous work where support for gender equality reduced when masculinity was threatened in an experimental setting (Kosakowska-Berzecka et al., 2016).

Analogously to gender identification, the extent to which an individual felt relatively young or old, based on their real age, was also taken into consideration. The measure of gender identification did not affect the extent or direction of stereotypes. However, those who felt younger rated their performance higher. So, even though there is no clear stereotype, this effect is in agreement with actual performance levels, as higher age negatively impacts performance.

It should be noted that the sample of participants of the current study was exclusively Dutch and thus may be entail some cultural restrictions. Furthermore, the Netherlands ranks relatively high on the 2020 Gender Equality Index (5th of 29 EU countries) (www.eige.europa.eu). The sample was of sufficient size to allow for a heterogeneous group within this cultural group, with variation in age and educational background.

To conclude, we found that for this large sample of Dutch participants, gender stereotypes concerning spatial abilities are very strong and held most strongly by male and young individuals. This age and gender related stereotype holds a direct relation to the patterns found for self-reported performance for gender, even though this is not supported by objective performance measures. Despite an increasing interest in aging effects in spatial cognition in the literature, age stereotypes are not held strongly. The lack of stereotypes impacts the self-reports in the opposite direction. As spatial abilities are engrained in many daily life activities, including education, implications of strongly held stereotypes can be considerable. As stereotype beliefs appear to differ across gender, males and females of any age may need to be addressed differently in order to counteract harmful effects of stereotypes. Age in particular may be relevant to consider in clinically relevant settings, such as self-reported cognitive performance in neuropsychological diagnostics and rehabilitation.