Seven Questions about Creativity and Creative Thinking

5. Thinking outside or inside the box? Patterns in idea flexibility and fixations in problem-solving tasks

In the context of measuring creativity and creative thinking, ideational fluency (i.e. being able to think of many ideas) and ideational flexibility (i.e. being able to think of very different ideas) are often used as proxies of creative abilities in divergent thinking tasks. Chapter 2 of this report focused on unpacking differences in the creative success of students in divergent and convergent thinking tasks included in the PISA CT Rescoring study, finding that students were more likely to propose creative ideas in the divergent thinking tasks in the study, on average across country-language. While students may ultimately have been more successful at thinking creatively in the divergent thinking tasks in the study, over a third of students across country-language groups were not able to come up with sufficiently different ideas according to judges in these tasks (two or three, depending on the task requirements) (Table B5.1).

One of the reasons individuals may find it difficult to think flexibly stems from a phenomenon known as “fixation effect”, which refers to the idea generation process being overly constrained to existing or obvious “known” solutions and thus limiting the generation of new and alternative solutions (Smith, 1995[1]). Fixation effects have been well documented in research on design thinking (Purcell and Gero, 1996[2]; Jansson and Smith, 1991[3]; Viswanathan and Linsey, 2011[4]), and research in cognitive psychology has also identified how various factors might induce fixation effects in creative tasks. These include through functional priming (Adamson, 1952[5]; Defeyter and German, 2003[6]) or activating individuals’ knowledge and memory by exposing them to existing solution examples or others’ ideas (Smith, Ward and Schumacher, 1993[7]; Agogué, 2015[8]). Studies have demonstrated that the type of examples shown to individuals can serve to strengthen or mitigate fixation effects (Agogué et al., 2014[9]), and that the age and expertise of individuals may determine how these factors interact with fixation effects (Cassotti et al., 2016[10]; Defeyter and German, 2003[6]; Viswanathan and Linsey, 2011[4]).

While the judges involved in the PISA CT Rescoring project awarded student responses a score for ideational flexibility as part of the criteria-based scoring method (see Annex A2 for more information), evaluating whether ideas are ‘sufficiently’ different or not in a cross-culturally valid way is challenging without a comprehensive rubric. Several of the studies cited above that focus on examining fixation effects employ a Concept-Knowledge (C-K) schema (Hatchuel and Weil, 2009[11]) to investigate the extent to which individuals can think flexibly and avoid idea fixations in divergent thinking tasks. The C-K approach involves creating an exhaustive mapping of combinations of concepts and knowledge within a given problem and solution space (see Box 5.1). This objective mapping, often conceptualised as a C-K ‘tree’ with branches, sub-branches and nodes, allows judges to map ideas onto a well-defined schema from which solution pathways can be inferred, as well as more granular idea flexibility and originality metrics.

We applied the C-K approach to two of the divergent thinking, problem-solving tasks included in the PISA CT Rescoring study: one relatively familiar social problem-solving task (Task 5 – Food Waste) and one more difficult, engineering-type task (Task 7 – The Exhibit). To what extent can students really ‘think outside of the box’ in different problem-solving tasks? Are different student groups more susceptible to fixation effects? Do we observe strong socio-linguistic differences in the types of ideas that students think of, or are students’ ideas more similar than we might expect across diverse socio-cultural and linguistic contexts?

5.1. How diverse are students’ ideas in problem-solving tasks?

5.1.1. Food Waste: a social problem-solving task in a familiar context

Task 5 in the PISA CT Rescoring study – Food Waste – was classified as a social problem-solving task in the PISA 2022 Creative Thinking assessment (Figure 5.1). The task presents students with the issue of food wastage in supermarkets and asks them to propose three different ways to address this problem. The problem context is relatively familiar and accessible: many if not all students that sat the PISA 2022 test would be familiar with supermarkets or grocers’ shops that sell fresh food items, and citizens including young adults in many countries have become increasingly socially engaged with issues related to reducing waste, recycling, etc. over the past decade. Students were generally successful at thinking of “sufficiently different” ideas for this task, according to judges’ criteria-based scores (see Box 5.1), with 65% of students on average across country-language groups achieving the maximum idea flexibility score and just 14% of all students unable to provide at least two different ideas (Table B5.2).

Data scored using the C-K Tree approach paint a much more nuanced story of the diversity of students’ ideas, both within and across country-language groups. The full C-K tree schema for Food Waste contained 8 primary solution paths (i.e. ‘branches’), with 25 sub-branches and 114 pre-defined end nodes in total (see Box 5.1 and Annexes C1 and C2). Of these 25 conceptually different sub-branches, just under 43% of student ideas on average across country-language groups were categorised within the same, single sub-branch: “Sub-branch E – distribute unsold produce for free” (Figure 5.2). In Chile and Portugal, over half of all ideas proposed by students corresponded to this sub-branch (Table B5.3). In some ways, the relatively high percentage of ideas within a single sub-branch is understandable, given that distributing unsold produce for free represents one of the most obvious solution paths for addressing the issue of food waste, and given that students were instructed to think of different ideas not necessarily creative ones. It therefore makes sense that many students chose an obvious solution for at least one of their three ideas.

Box 5.1. Which score data are used in this chapter?

The analyses in this chapter draw upon two types of rescored data from the PISA CT Rescoring study:

• Conceptual-Knowledge (C-K) Tree codes: These data are categorical and describe, for each idea within a student response, its corresponding sub-branch and node on the task-specific Conceptual-Knowledge tree. The process for developing the C-K trees for the two tasks discussed in this chapter is described in Annex A2, and the full C-K tree schema and accompanying list of nodes for both tasks are detailed in Annexes C1 to C4.

• Criteria-based scores: Some analyses draw upon response flexibility scores attributed by judges based on the number of “sufficiently different” ideas provided by the student within a response.

This chapter focuses on examining ideational flexibility and identifying fixation paths in student responses to two problem-solving tasks across country-language groups.

Figure 5.1. Food Waste task

Source: PISA 2022 Creative Thinking test.

Perhaps more surprisingly than the large share of ideas across country-language groups within the same sub-branch is that, even amongst all ideas within that sub-branch, students were remarkably consistent in their ideation: around 4 in 10 ideas on average across all country-language groups corresponded to the same specific node of the C-K Tree (“give produce to people directly for free”) (Table B5.5). Only in Canada (English provinces) and Saudi Arabia did students think of slightly different ideas within that same sub-branch most frequently (“donate produce to charities”).

Table 5.1 summarises the 5 most common idea nodes across the full C-K Tree schema across country-language groups in the study. Overall, of the top 5 most common ideas (i.e. nodes), three were situated within sub-branch E: “give produce directly to people for free” (node ee; 23% ideas on average), “donate produce to charities” (node ea; 12% ideas on average) and “donate produce to animals or animal/agricultural structures” (node ei; 7% ideas on average). Despite the C-K Tree schema defining 114 end nodes a priori, the 5 most common nodes account for 62% of all ideas. However, only the 3 most common nodes on average across country-language groups account for more than 10% of all ideas each, with the rest of students’ ideas spread across a more diverse range of nodes.

Figure 5.2. Nearly half of all student ideas in Food Waste fall within the same sub-branch of the C-K tree

Percentage of student ideas within each branch and select sub-branches, on average across country-language groups

Note: Only sub-branches with over 2% of all student ideas on average across country-language groups are shown. Percentages shown in the figure are rounded to the nearest integer. The full C-K Tree schema for this task is detailed in Annex C1.

Source: Table B5.3

Table 5.1. Three out of the five most common ideas in Food Waste fall within the same sub-branch

Percentage of all student ideas within the 5 most common nodes (on average across country-language groups) on the C-K Tree for Food Waste

Node description	%	Sub-branch (node)
Give produce to people directly for free	23%	E (ee)
Lower the price of individual products	13%	I (ia)
Donate to charities	12%	E (ea)
Donate to animals or agricultural/animal structures	7%	E (ei)
Supermarkets to buy fewer products	7%	C (ca)

Note: Percentages are rounded to the nearest integer. For a full description of all nodes for this task, see Annex C2.

Source: Table B5.6

Beyond the sub-branch “distribute unsold produce for free”, only two other sub-branches amongst all those available in the problem space contain more than 10% of all student ideas each, on average across country-language groups. These are “sell products cheaper” (sub-branch I) and “reduce the quantity of products purchased by supermarkets” (sub-branch C), respectively. Table 5.2 describes the 5 sub-branches with the largest share of ideas, on average across country-language groups.

Table 5.2. Over two thirds of all ideas focus on addressing the problem of food waste once products are already in the supermarket

Percentage of all student ideas within the 5 most common sub-branches (on average across country-language groups) on the C-K Tree for Food Waste

Sub-branch description	%	Branch
E – Distribute unsold produce for free	43%	Act once products are in store 🡺 Act on storage 🡺 Empty stock
I – Sell products cheaper	13%	Act once products are in store 🡺 Through sales strategies 🡺 Target products
C – Reduce the quantity of products purchased by supermarkets	10%	Act before products arrive in store
U – Recycle/transform waste into non-edible product	9%	Act once products are in store 🡺 Transform product
S – Transform waste food into another food product	4%	Act once products are in store 🡺 Transform product

Note: Percentages are rounded to the nearest integer. For a full presentation of the C-K schema for this task, including all sub-branches, see Annex C1.

Source: Table B5.3

Figure 5.3 shows the relative share of ideas within these 5 sub-branches in select country-language groups. While the relative shares vary within each country-language group, three things are notable. First, sub-branch E ("distribute unsold produce for free") represents the most common sub-branch in all country-language groups by a significant margin. Second, beyond sub-branch E, the distribution of ideas across the remaining sub-branches is relatively stable across country-language groups, with the cross-cultural variation remaining within 5 percentage points of the country-language average percentage in most cases (Table B5.3). This relative stability remains even when limiting the analysis to appropriate ideas only (i.e. those that achieve at least a score of 1 point for appropriateness), with only small changes to the overall distribution of ideas across all sub-branches (Table B5.4). Third, although the relative ordering of the sub-branches varies in each country-language group included in the study, particularly those that are second to fifth most common, it is notable that the five most common within-country idea nodes always correspond to one of these 5 sub-branches (Table B5.3 and Table B5.6). What these data show is that despite significant socio-cultural and linguistic diversity amongst students in the PISA CT Rescoring project sample, there is nonetheless broad homogeneity in the types of ideas that students suggest for this task.

Figure 5.3. Cross-cultural variability is relatively stable in Food Waste

Percentage of ideas within the 5 most common sub-branches (on average across country-language groups), by select country-language groups

Note: The country-language groups with the largest and smallest shares of ideas within each of the 5 sub-branches are shown in the Figure, though not all country-language groups with available data are shown. Country-language groups are shown in alphabetical order following the average. The top 5 sub-branches are ordered by share of ideas at the average country-language level (highest to lowest). The max-min difference shows the percentage difference in the share of ideas between the country-language group with the highest share of ideas and the country-language group with the lowest share of ideas within each sub-branch. Percentages are rounded to the nearest integer.

Source: Table B5.3

Table 5.2 also describes the primary solution paths (or broader branch of the C-K Tree) that lead to the 5 most common sub-branches, on average across country-language groups. Notably, four of the five most common sub-branches sit within the same major branch of the C-K Tree (“act once fresh products are in the store”); this branch represents around 83% of all student ideas, on average across country-language groups, though it also encompasses several primary solution paths and the vast majority of the defined sub-branches within the solution space, including introducing sales strategies, transforming products into other edible or non-edible products, or acting on the storage of products by emptying stock or improving storage conditions (Figure 5.2). In contrast, around 10% of ideas corresponding to sub-branch C (“reduce the quantity of products purchased by supermarkets”) – the third most frequently suggested sub-branch, on average – suggest addressing the problem from an alternative angle and solution path, that is taking action before the fresh products arrive at supermarkets. Very few students on average across country-language groups consider addressing the problem through other primary solution paths such as introducing some sort of regulatory framework or policies (around 3% ideas) or by reimagining the constraints of the problem itself (less than 1%).

Overall, these distributions of ideas across country-language groups for the Food Waste task imply two things that, at first, may seem contradictory. First, the high concentration of ideas within a select few solution paths, sub-branches and even specific nodes suggests that many students, across many country-language groups, think similarly when asked to come up with different ideas for addressing the problem of food waste in supermarkets. Second, the data suggest that beyond these few obvious solution ideas, students’ ideas become quite varied. Indeed, 10 of the 25 sub-branches in the full C-K Tree schema contained between 1-10% of all ideas each, with a further 3% of ideas remaining unattributed to an existing category (Table B5.3).

These two phenomena may co-exist given the characteristics of the Food Waste task: there is a relatively large problem and solution space for students to explore; and the task is broadly accessible and familiar to students in some way, meaning both that students will have a shared knowledge about broad solution approaches already implemented by supermarkets across contexts (e.g. giving away food at the end of the day or reducing prices), but that students may also have been directly exposed to various different or culturally-specific approaches in their own socio-cultural context or lived experience. For example, while sub-branch E (“distributing unsold produce for free”) was consistently and significantly the most common sub-branch of solution ideas in all country-language groups, with “giving food directly to others for free” the most common instantiation (i.e. node) of this group of ideas in most country-language groups (around 23% of all ideas, on average), in Saudi Arabia, it was not even in the top 5 most common nodes; instead, a sizeable proportion of students (11%) in Saudi Arabia referred to creating a lottery system where people could win food (Table B5.6). Thus, while the broad solution approach remains similar, differences in the specific implementation of ideas might reflect differences in socio-cultural norms, values and practices.

5.1.2. The Exhibit: a harder scientific problem-solving task

Do we observe a similar diversity and spread of ideas in a different problem-solving task included in the PISA CT Rescoring study? The C-K Tree scoring method was also applied to Task 7 – The Exhibit. This task was classified as a scientific problem-solving task in the PISA 2022 Creative Thinking test as it combines elements of physics, biology and engineering design. The task asks students to propose design ideas for constructing a separating wall in an animal park exhibit that would allow squirrels to pass from one room to the other for feeding purposes, while preventing rats in the enclosure from doing so (Figure 5.4). The item specified that while the squirrels and the rats have the same body size, the squirrels are heavier, faster and more agile than the rats.

The Exhibit is a comparatively harder task than Food Waste. In The Exhibit, students must combine their understanding of the characteristics of the two animals, as described in the task scenario, with some basic understanding of physics and engineering principles. It thus is more demanding from a domain knowledge point of view, and, unlike the Food Waste task, students may have limited lived experience with or exposure to such problems. The Exhibit also generally has a higher cognitive load for students (more text to read and understand, and greater demands on the level of detail required in responses). Indeed, The Exhibit appeared the most difficult of the seven items included in the PISA CT Rescoring project in terms of student success in generating creative ideas, with a mean idea sum score of just 2.8 score points (on average across country-language groups) compared to a mean idea sum score of 3.7 score points on average across all tasks (Table B1.17 and Table B2.2). The criteria-based scores for response flexibility also indicate that students found it relatively difficult to generate multiple different ideas for this task, with 45% of students achieving a score of 0 for response flexibility and fewer than 1 in 3 students able to provide two sufficiently different ideas (Table B5.7).

The relative difficulty of The Exhibit task compared to the Food Waste task can also be viewed through the comparative openness of the problem and solution spaces. Although the C-K Tree for The Exhibit has 7 primary solution paths (branches), compared to the 8 primary solution paths in Food Waste, each path is less expansive: the full C-K Tree schema for The Exhibit includes just 13 sub-branches and 57 pre-defined end nodes, roughly half of those that were included in the equivalent for Food Waste (see Annexes C1 to C4).

Figure 5.4. The Exhibit task

Source: PISA 2022 Creative Thinking test.

Figure 5.5 shows that the majority of student ideas across country-language groups are split between two sub-branches of the C-K Tree in The Exhibit: creating an opening that allows squirrels to pass through “due to their speed/agility” (34% all ideas; sub-branch B) or “due to their morphology” (25% all ideas; sub-branch A). Like in Food Waste, it is unsurprising that a significant share of all ideas corresponds to the same sub-branches given the task asks students to come up with different ideas of any quality. It could also be expected that the two most common ideas relate to characteristics of the squirrels that are directly referenced in the task prompt, given prior research on the effect of activating knowledge and providing examples on constraining flexibility in divergent thinking tasks (Agogué et al., 2014[9]; Smith, Ward and Schumacher, 1993[7]).

Figure 5.5. Over half of all ideas in The Exhibit correspond to two sub-branches of the C-K Tree

Percentage of student ideas within each branch and select sub-branches, on average across country-language groups

Note: Only sub-branches with over 2% of all student ideas on average across country-language groups are shown. Percentages shown in the figure are rounded to the nearest integer. The full C-K Tree schema for this task is detailed in Annex C3.

Source: Table B5.8

One difference in the distribution of ideas in The Exhibit task with respect to the Food Waste task is that ideas are relatively more evenly distributed across the different sub-branches and nodes that populate the solution space. Most ideas in the Food Waste task are captured by just a few individual nodes, with a large spread of ideas across the remaining sub-branches and nodes in the solution space. In contrast in The Exhibit, student ideas were less concentrated within a specific node – both across and within sub-branches. For example, Table 5.3 shows that although the 5 most common nodes in The Exhibit task account for a similar proportion of all ideas as in the Food Waste task (between 61-62%), student ideas are more evenly distributed across these 5 nodes in The Exhibit. Table 5.4 also shows how the modal nodes by country-language group vary within the two most popular sub-branches (on average).

Table 5.3. Three out of the five most common ideas in The Exhibit focus on designs related to the speed or agility of the squirrels

Percentage of all student ideas within the 5 most common nodes (on average across country-language groups) on the C-K Tree for The Exhibit

Node description	%	Sub-branch (node)
Create an elevated path or course for the squirrel to traverse to the other side	16%	B (bf)
Separate the rats and squirrels permanently (i.e. avoid the problem altogether)	13%	M (mc)
Create an elevated opening that only the squirrel can jump to	13%	B (be)
Create an opening that opens with a balance/pressure plate on which the squirrel positions itself	11%	A (ae)
The opening is accessible only with the speed of the squirrel	8%	B (bb)

Note: Percentages are rounded to the nearest integer. For a full description of all nodes for this task, see Annex C4.

Source: Table B5.10

Table 5.4. The most common nodes within sub-branches A and B vary significantly by country-language group

Modal nodes by country-language group, within the two sub-branches with the largest share of student ideas (on average across country-language groups)

Sub-branch A: An opening for the squirrel based on its morphology			Sub-branch B: An opening for the squirrel based on its speed/agility
Modal node	Node description	Country-language groups	Modal node	Node description	Country-language groups
ab	Opening is a small hole that only squirrels can pass through	Saudi Arabia	bb	Opening only accessible by speed	Hong Kong (China, Chinese speaking), Korea
ac	Make a tunnel opening	Colombia	be	Create an elevated opening (squirrels must jump to access)	Brazil, Canada (English provinces), France, Greece, Saudi Arabia
ad	An object (e.g. door) that only squirrels have the strength to push past	Hong Kong (China, English speaking)
ae	A balance/pressure plate that recognises squirrels	Belgium (French community), Brazil, Canada (English and French provinces), Chile, France, Greece, Hong Kong (China, Chinese speaking), Korea, Slovenia	bf	Create an elevated path that squirrels must traverse to the other side	Belgium (French community), Canada (French provinces), Chile, Colombia, Germany, Hong Kong (China, English speaking), Slovenia
ag	Counterweight system	Germany

Note: For a full description of all nodes for this task, see Annex C4.

Source: Table B5.11

In general, the cross-cultural variability in the distribution of ideas across sub-branches is greater in The Exhibit task compared to the Food Waste task. Figure 5.6 shows the share of ideas in select country-language groups within the 5 sub-branches that contain the largest shares of ideas (on average across country-language groups). The percentage of ideas varies significantly within sub-branch B (“creating an opening that allows squirrels to pass through due to their speed/agility”), ranging from 22% of all ideas in Greece to 52% ideas in Germany (Table B5.8). Although sub-branch B remains the most common sub-branch in 13 of the 16 country-language groups, the most common idea node in around half of the country-language groups corresponds to a different sub-branch (sub-branch A “an opening that only squirrels can pass through due to their size” or M “modify the spatial-temporal characteristics of the problem”) (Table B5.11). In general, there is significant variation across country-language groups in the relative shares of ideas in four of the five most common sub-branches (A, B, M and H), with the relative shares of ideas in several country-language groups exceeding 5 percentage points of the country-language average in several instances (Table B5.8). The distribution of ideas across sub-branches also changes significantly when limiting the analysis to appropriate ideas only (i.e. those that achieve at least a score of 1 point for appropriateness), with a much greater concentration of ideas within sub-branches A and B on average across country-language groups (Table B5.9).

Figure 5.6. There is significant cross-cultural variability in the share of ideas across sub-branches in The Exhibit

Percentage of ideas within the top 5 sub-branches (on average), by select country-language groups

Note: The country-language groups with the largest and smallest shares of ideas within each of the 5 sub-branches are shown in the Figure, though not all country-language groups with available data are shown. Country-language groups are shown in alphabetical order following the average. The top 5 sub-branches are ordered by share of ideas at the average country-language level (highest to lowest). The max-min difference shows the percentage difference in the share of ideas between the country-language group with the highest share of ideas and the country-language group with the lowest share of ideas within each sub-branch. Percentages are rounded to the nearest integer.

Source: Table B5.8

The comparative difficulty of The Exhibit task may influence these differences in the variation and distribution of ideas, likely because proposing functional solutions requires some degree of relevant knowledge from a range of disciplines. These differences in domain readiness across country-language groups may be reflected in the larger variation in modal nodes in The Exhibit task as well as the overall greater cross-cultural variation in the distribution of students’ ideas. However, when considering only appropriate ideas suggested by students, the distribution of ideas is much more concentrated within the two sub-branches associated with the characteristics of the animals mentioned explicitly in the task prompt.

A sizeable proportion of students in The Exhibit task proposed ideas that significantly changed the problem presented (around 13%), for example by simply separating the two animals definitively and thus getting rid of the problem altogether (Table B5.8). In contrast, only 0.2% students proposed a similar problem-avoiding solution in the Food Waste task (Table B5.3). It may be that the harder difficulty of the problem, or the relative lack of on-topic solution ideas, caused students to think about ways to reframe the problem. While these types of “solutions” may well be original in the sense of adopting alternative perspectives, their appropriateness is questionable. However, in certain real-life situations, such lateral thinking and alternative perspective-taking may be useful.

The data from the C-K Trees also confirms that students found it relatively harder to think of different ideas in The Exhibit task. Figure 5.7 shows the percentages of students that suggested ideas in different sub-branches of the C-K Trees for both tasks. Although similar proportions of students were able to think of ideas corresponding to two different sub-branches of the C-K Tree in each task, many more students in The Exhibit task suggested multiple ideas within the same sub-branch compared to the same in the Food Waste task (30% students compared to 16%).

Figure 5.7. Students found it harder to suggest ideas in different sub-branches in The Exhibit than in Food Waste

Share of students that proposed ideas in one, two or three different sub-branches of the C-K Tree for the Food Waste and The Exhibit tasks (on average across country-language groups)

Note: Percentages are rounded to the nearest integer.

Source: Table B5.12 and Table B5.13

5.2. Are there differences in idea fixations across different student groups and socio-linguistic contexts?

For both the Food Waste and The Exhibit tasks, it is possible to define “fixation paths” and “expansion paths” (Box 5.2). Fixation paths refer to solutions mapped onto primary solution paths that many students follow, whereas expansion paths refer to solutions proposed outside of these most-followed solution paths. For the analyses in the rest of this chapter, we examine whether there are differences in the degree to which students generate ideas within fixation or expansion paths across different student groups and different socio-linguistic contexts.

Box 5.2. Identifying international fixation paths in students’ thinking

In previous research using the C-K method, “fixation paths” have been defined in the dataset as those that account for around 80% of all ideas in the sample (Agogué et al., 2014[9]).The fixation and expansion paths referred to in this chapter have been defined with respect to the international average, i.e. the sub-branches that account for the majority of all student ideas on each task, on average across country-language groups. Given the comparative distribution of the data across our two tasks, the sub-branches included in the fixation paths are those that cumulatively, from largest share to smallest share, include at least 79% of all student ideas on average across country-language groups. This represents the 5 most common sub-branches in Food Waste and the 4 most common sub-branches in The Exhibit.

Table 5.5. Sub-branches and cumulative frequency of ideas within the international fixation path for Food Waste and The Exhibit

Task 5 – Food Waste				Task 7 – The Exhibit
#	Code	Sub-branch	Cum. Freq. (%)	#	Code	Sub-branch	Cum. Freq. (%)
1	E	Distribute unsold produce for free	43	1	B	Create an opening that squirrels can pass through due to their speed/agility	34
2	I	Sell products cheaper	55
3	C	Reduce the quantity of products purchased by supermarkets	65	2	A	Create an opening that squirrels can pass through due to their morphology	60
4	U	Recycle/transform waste into non-edible product	74	3	M	Modify the spatial-temporal characteristics of the problem	72
5	S	Transform waste into another food product	79	4	D	Create an opening that recognises the squirrels	79

Source: Table B5.14

Figure 5.8. Share of ideas in sub-branches within the international fixation path for Food Waste and The Exhibit

Note: Sub-branches are shown in descending order of the share of ideas at the average country-language group level.

Source: Table B5.3 and Table B5.8.

5.2.1. By gender and socio-economic status

Our data show that there are significant differences in idea fixation across student groups. Figure 5.9 shows that, in both tasks, girls are significantly relatively more likely than boys to propose solution ideas within the fixation path, on average across country-language groups. Advantaged students are also relatively more likely than disadvantaged students to suggest solution ideas within the fixation path, on average across country-language groups, though differences between advantaged and disadvantaged students are not significant in the Food Waste task. These patterns regarding relative differences in the ideation paths by gender and socio-economic status remain consistent even when replicating the analyses using nationally referenced fixation paths for each country-language group rather than the international fixation path, though differences between student groups tend to be smaller (Tables B5.15-B5.18).

These findings might seem counter-intuitive at first glance: how is it that girls and advantaged students, who performed better on the PISA 2022 creative thinking assessment (OECD, 2024[12]) and who tended to outscore boys and disadvantaged students in the PISA CT Rescoring project (see Chapters 3, 6 and 7), also propose more ideas corresponding to the fixation path? What these results suggest is that creative ideas are not necessarily those that are “outside of the box” – at least, not in the types of everyday problem-solving tasks included in the PISA assessment. The PISA assessment tasks aimed to elicit “little c” expressions of creativity, rather than the types of “Big C” creative advances that might be more associated with out-of-the-box thinking. Ivcecic and Kaufman (2024[13]) underline the importance of task relevance for achieving “little-c” creative outcomes. In this way, it may be that “inside the box” ideas that capture some elements of originality are those that are most likely to be considered creative, rather than very remote or original ideas that only partially address the task.

Figure 5.9. Girls and advantaged students are more likely to suggest ideas within the fixation path than boys and disadvantaged students

Relative percentage of ideas in the within-task fixation path (on average across country-language groups), by gender and socio-economic status

Note: Significant differences between student groups are shown by filled markers. The relative percentage of ideas refers to the share of ideas amongst all student ideas within a given student group (i.e. amongst all girls, the percentage of ideas that correspond to sub-branches within the fixation path). The differences shown between student groups refer to the differences in relative percentages of student ideas within the fixation path and are rounded to the nearest integer. The fixation path is defined with respect to the international average (see Box 5.2).

Source: Tables B5.15-B5.18.

In both tasks, the sub-branches that contained the highest scoring ideas (on average) in country-language groups were not necessarily those that might be considered the most original types of ideas (based on frequency) amongst all the possibilities defined within the C-K Tree schema; however, they all clearly respond to the task goals in relatively practical ways (Table B5.19 and Table B5.20). For example, in The Exhibit task, sub-branches D (“an opening that recognises squirrels”), A (“an opening that only squirrels can pass through due to their morphology”) and B (“an opening that only squirrels can pass through due to their speed/agility”) are the modal sub-branches containing the highest mean idea sum scores (i.e. as either the highest-scoring or second-highest scoring sub-branch) within each country-language group. In the Food Waste task, the equivalent modal sub-branches containing the highest mean idea sum scores within each country-language group are sub-branches C (“reduce the quantity of products purchased by supermarkets”), U (“recycle or transform waste into non-edible product”) or S (“transform the produce into another food product”). In both tasks, these modal, high-scoring sub-branches all fall within the respective solution fixation paths.

Girls and advantaged students may be more likely to search for creative ideas “inside the box” for two reasons. Firstly, these student groups may be more willing or disposed to pay attention to and comply with the task goals, compared to boys and disadvantaged students, respectively. The analyses elsewhere in this report show that girls are significantly more likely to suggest appropriate ideas than boys, in general (see Chapter 6), and the PISA 2022 results found that boys tended to show more disengaged task behaviours than girls (OECD, 2024[12]). Not only do the fixation path sub-branches tend to correspond to appropriate classes of ideas to begin with, but girls also propose significantly more appropriate ideas than boys even within the fixation path in Food Waste, on average across country-language groups (Table B5.21). Similarly, advantaged students also propose significantly more appropriate ideas than disadvantaged students within the fixation path for both tasks, on average across country-language groups (Table B5.23 and Table B5.24).

Secondly, being able to provide more effective and relevant solutions inevitably requires domain-relevant knowledge. Both girls and advantaged students tend to outperform boys and disadvantaged students respectively in PISA, in general, suggesting that these latter groups may have more difficulty in understanding the task requirements or providing relevant, original solutions of quality in creative thinking tasks. Again, even amongst ideas within the fixation paths (so, those that are relatively common), advantaged students propose significantly more original and valuable ideas than their disadvantaged peers on average across country-language groups (Table B5.23 and Table B5.24). While there are no significant differences in the originality and value of ideas specifically within the fixation path between boys and girls, girls’ overall idea sum scores are nonetheless significantly higher than boys’ on average across country-language groups in both tasks (Table B5.21 and Table B5.22).

5.2.2. By socio-linguistic context

What about differences in idea fixations across socio-linguistic contexts? To what extent are our ideas and thinking patterns shaped by our cultural experiences and the language that we speak? Figure 5.10 and Figure 5.11 show the percentage of students’ ideas that fall within the international fixation path for the two tasks, respectively, by country-language group, as well as by language group and geographic region (where these include at least two country-language groups that participated in the PISA CT Rescoring project).

As expected, the degree to which student ideas fall within the fixation path varies across different socio-linguistic groups. However, it’s also notable that in most country-language groups, the fixation path does account for close to three quarters or more of all student ideas – again suggesting that student ideas across country-language groups for these two tasks are remarkably consistent. Of course, the fixation paths refer only to the sub-branches of solution pathways and not to specific nodes on the C-K trees, where there is larger variation in students’ ideas, but it is striking that most students in most country-language groups follow the same general solution pathways. Part of this effect might be due to the nature of the PISA 2022 divergent thinking tasks – asking for students to generate different ideas of any quality, rather than focusing on coming up with different creative ideas – but the analyses in Chapter 2 of this report also demonstrated that students generally produced more creative ideas as a byproduct in these divergent thinking task types.

Figure 5.10. At least two thirds of ideas within each country-language group corresponds to the international fixation path for the Food Waste task, though the share varies significantly

Percentage of student ideas within the international fixation path for Food Waste, by country-language group, language group and geographic region group

Note: Percentages are rounded to the nearest integer. Geographic region averages include at least two country-language groups. The sample size for Hong Kong (China, English speaking) is very small, so results for this country-language group should be interpreted with caution (for more information, see Annex A1). The country-language average is computed including Hong Kong (China, all), not Hong Kong (China, English speaking) and Hong Kong (China, Chinese speaking) as separate entities.

Source: Table B5.25

Figure 5.11. At least two thirds of ideas within each country-language group corresponds to the international fixation path for The Exhibit task, though the share varies significantly

Percentage of student ideas within the international fixation path for Food Waste, by country-language group, language group and geographic region group

Note: Percentages are rounded to the nearest integer. Geographic region averages include at least two country-language groups. The sample size for Hong Kong (China, English speaking) is very small, so results for this country-language group should be interpreted with caution (for more information, see Annex A1). The country-language average is computed including Hong Kong (China, all), not Hong Kong (China, English speaking) and Hong Kong (China, Chinese speaking) as separate entities. There are no data for Portugal for this task.

Source: Table B5.26

In both tasks, there are some relative outlier country-language groups. In Saudi Arabia, only around 67% of all student ideas fell within the international fixation path in the Food Waste task. Students were significantly more likely to suggest ideas related to the production of fresh food (i.e. grow less food in general or modify the type of food grown) and to the storage of fresh goods in supermarkets than on average across country-language groups in this task, and they were also significantly less likely to suggest ideas related to (re)distributing fresh food for free or transforming fresh food waste into other (non-consumable) products. Relatively fewer ideas for The Exhibit task were also captured by the fixation path for students in Saudi Arabia compared to the country-language average (difference of 8 percentage points).

In Hong Kong (China, English speaking), only between 68-75% of all student ideas are within the fixation path for both tasks. These results may be an artefact of the small student sample size for the English speaking student population in the sample; however, the Chinese-speaking sample of students from Hong Kong (China) also suggest fewer ideas within the fixation path than in most other countries, in both tasks. These results imply that students in Hong Kong (China) are particularly good at thinking flexibly in problem-solving tasks, or that there may be strong cultural differences in the primary solution paths that students in Hong Kong (China) follow compared to other country-language groups.

Key findings and implications

• In general, students do not think much outside of the box when asked to come up with different ideas to everyday problem-solving tasks – both within and across country-language groups

  • In both problem-solving tasks, many students in many countries follow relatively fixed solution paths. In the Food Waste task, over 4 in 10 ideas corresponded to the same sub-branch of the C-K Tree, on average, and in both tasks, the 5 most common nodes accounted for over 60% of all student ideas on average across country-language groups.

• There was greater cross-cultural variability in students’ ideas in The Exhibit task compared to the Food Waste task, with ideas more evenly spread across nodes (in general) and greater variability in the relative shares of ideas within different sub-branches

  • Part of this greater cross-cultural variability may be due to the task difficulty and differences in students’ prior knowledge and experience across relevant disciplines (e.g. biology, physics, engineering).

• Knowledge matters in supporting creative problem-solving: it is important to teach students how to apply creative thinking in domain-relevant ways, enabling them to find appropriate and practical solutions with an original twist

  • Students were far more likely to propose ideas that ignored the task requirements or that avoided the problem altogether in The Exhibit, again perhaps due to the greater difficulty of the task and a relative lack of familiar solution approaches from which to suggest appropriate ideas.

• It may neither be reasonable nor desirable to expect students to think outside of the box in these types of everyday problem-solving contexts: successful “everyday” creativity might be more about thinking inside the box in creative ways (Ivcevic and Kaufman, 2024[13])

  • Girls and advantaged students seem to be more successful at thinking creatively “inside the box”, most likely due to differences across student groups in task engagement, domain-relevant knowledge, and being able to identify more appropriate solutions.

References

[5] Adamson, R. (1952), “Functional fixedness as related to problem solving: A repetition of three experiments.”, Journal of Experimental Psychology, Vol. 44, pp. 288-291.

[8] Agogué, M. (2015), “Resisting classical solutions: The creative mind of industrial designers and engineers”, Psychology of Aesthetics, Creativity, and the Arts, Vol. 9/3, pp. 313-318.

[9] Agogué, M. et al. (2014), “The impact of type of examples on originality: explaining fixation and stimulation effects”, Journal of Creative Behavior, Vol. 48/1, pp. 1-12, https://doi.org/10.1002/jocb.37.

[10] Cassotti, M. et al. (2016), “Fixation effect in creative ideas generation: Opposite impacts of example in children and adults”, Thinking Skills and Creativity, Vol. 19, pp. 146-152, https://doi.org/10.1016/j.tsc.2015.10.008.

[6] Defeyter, M. and T. German (2003), “Acquiring an understanding of design: evidence from children’s insight problem solving”, Cognition, Vol. 89/2, pp. 133-155, https://doi.org/10.1016/s0010-0277(03)00098-2.

[11] Hatchuel, A. and B. Weil (2009), “CK design theory: An advanced formulation.”, Research in Engineering Design, Vol. 19, pp. 181-192.

[13] Ivcevic, Z. and J. Kaufman (2024), “Inside the box: Considering the role of appropriateness for creativity”, Psychology of Aesthetics, Creativity and the Arts, https://doi.org/10.1037/aca0000703.

[3] Jansson, D. and S. Smith (1991), “Design fixation”, Design Studies, Vol. 12/1, pp. 3-11, https://doi.org/10.1016/0142-694X(91)90003-F.

[12] OECD (2024), PISA 2022 Results (Volume III): Creative Minds, Creative Schools, PISA, OECD Publishing, https://doi.org/10.1787/765ee8c2-en.

[2] Purcell, A. and J. Gero (1996), “Design and other types of fixation”, Design Studies, Vol. 17/4, pp. 363-383, https://doi.org/10.1016/S0142-694X(96)00023-3.

[1] Smith, S., T. Ward and R. Finke (eds.) (1995), Fixation, incubation and insight in memory, problem-solving, and creativity, MIT Press.

[7] Smith, S., T. Ward and J. Schumacher (1993), “Constraining effects of examples in a creative generation task”, Memory and Cognition, Vol. 21, pp. 837-845.

[4] Viswanathan, V. and J. Linsey (2011), Understanding fixation: A study on the role of expertise.