Neurobiological origins

The emergence of CF is linked to maturation of the prefrontal cortex (PFC) and inferior parietal cortex (Ezekiel, Bosma, & Morton, Reference Ezekiel, Bosma and Morton2013). Although the PFC is relatively late maturing, this region already begins to undergo synaptic pruning and myelination from the first year of life (Collin & van den Heuvel, Reference Collin and van den Heuvel2013). Between 1 and 2 years, there is a pronounced acceleration in the volume of prefrontal gray matter, with expansion in both cortical thickness and surface area (Gilmore et al., Reference Gilmore, Shi, Woolson, Knickmeyer, Short, Lin and Shen2012). During the early years, external environmental stimuli are important for shaping the developmental trajectory of the PFC (Chini & Hanganu-Opatz, Reference Chini and Hanganu-Opatz2021). Specifically, factors such as caregiver interactions (Nelson, Reference Nelson2007), pronounced sensory deprivation (McLaughlin, Sheridan, & Lambert, Reference McLaughlin, Sheridan and Lambert2014), prenatal exposure to substances (Mackey, Raizada, & Bunge, Reference Mackey, Raizada, Bunge, Bunge, Toga, Stuss and Knight2013), and early adverse experiences (Hodel, Reference Hodel2018) can impact PFC maturation. Neurotransmitter systems are involved in the neuromodulation of CF. For example, studies have shown that serotonin (Chamberlain et al., Reference Chamberlain, Muller, Blackwell, Clark, Robbins and Sahakian2006; Skandali et al., Reference Skandali, Rowe, Voon, Deakin, Cardinal, Cormack and Sahakian2018) and dopamine (Dang, Donde, Madison, O'Neil, & Jagust, Reference Dang, Donde, Madison, O'Neil and Jagust2012) both affect performance on CF tasks. In contrast noradrenaline does not seem to affect at least some CF tasks such as the CANTAB IED (Chamberlain et al., Reference Chamberlain, Muller, Blackwell, Clark, Robbins and Sahakian2006).

Measuring emerging cognitive flexibility

Precursors of CF such as attention set-shifting, reversal learning, and overcoming perseveration begin to emerge during infancy. Specifically, one can assess infants' ability to overcome perseveration and engage in reversal learning from as early as 6–12 months (de Sousa, de Gil, & McIlvane, Reference de Sousa, de Gil and McIlvane2015). The sequential touching task (Ellis & Oakes, Reference Ellis and Oakes2006) has been adapted as an infant CF measure to assess flexible attention set-shifting from 12 months of age, particularly when the shift is scaffolded by a social partner (Fig. 2a). Piaget's A-not-B task, a classic test of infant cognitive development, requires basic shifting, memory and inhibitory control skills, and children still make errors on this task until ~12 months of age (MacNeill, Ram, Bell, Fox, & Pérez-Edgar, Reference MacNeill, Ram, Bell, Fox and Pérez-Edgar2018). Together with other core EF skills, CF development accelerates during the preschool years (Hughes, Reference Hughes1998) and is only thought to reach maturity during late childhood or early adolescence (Kupis & Uddin, Reference Kupis and Uddin2023). In children, a more formal measure of rule switching is the Dimensional Change Card Sort task (DCCS, Fig. 2b), in which children are asked to sort cards according to one dimension (e.g. color, shape, or number) and this sorting role is changed after several trials. While 4-year-old children are typically able to switch successfully between different dimensions, 3-year-old children tend to perseverate on one dimension (Doebel & Zelazo, Reference Doebel and Zelazo2015). In the Intra-Extra Dimensional set shift task (IED, from Cambridge Neuropsychological Test Automated Battery), typically-developing 5-year-old children already display successful attentional set-shifting, but 7 to 18-year-old children with autism show significant CF dysfunction (Hughes et al., Reference Hughes, Russell and Robbins1994; Langley, Sahakian, & Robbins, Reference Langley, Sahakian, Robbins, Boyle, Stein, Stern, Sahakian, Golden, Lee and Chen2023). Childhood CF is linked to essential life outcomes such as social skills, learning, financial stability, and overall well-being (Arán Filippetti & Krumm, Reference Arán Filippetti and Krumm2020; Broomell & Bell, Reference Broomell and Bell2022).

Figure 2.

Examples of age-appropriate CF tasks for infants, children, and adults. (a) In the Sequential Touching and Object Categorization (STOC) task, infants are presented with objects that can be categorized by either a high-salience dimension (e.g. shape: balls v. blocks) or a low-salience dimension (e.g. material: soft v. hard). This task comprises three phases: phase 1- infant free play; phase 2- parent demonstration of toy material (compressibility); phase 3- infant free play. The STOC measures flexible attention set-shifting in infants' mental categorization of toy objects from 12 months of age, particularly when the shift is scaffolded by a social partner (Tan & Leong, Reference Tan and Leong2023). (b) In the Dimensional Change Card Sort (DCCS) task, children sort cards based on one dimension (e.g. color) and after several trials, they are instructed to switch and sort by another dimension (e.g. shape). The task assesses their ability to shift between different sets of rules and adapt to new instructions (Zelazo, Reference Zelazo2006). (c) In the Wisconsin Card Sort Task (WCST), participants sort a target card into one of four decks without knowing the initial sorting rule. After each sort, they receive feedback on its correctness. From this feedback, they must infer the underlying rule. After several consecutive correct responses, the sorting rule changes without notice, challenging participants to detect the shift and adjust their strategy accordingly (Tong et al., Reference Tong, Chan, Cheng, Cheon, Ellefson and Fauziana2023).

Adolescence and young adulthood

During adolescence, performance on CF tasks continues to improve in tandem with increasing brain specialization (Kupis & Uddin, Reference Kupis and Uddin2023). Although adolescent CF approximates adult levels from ~12 years (Huizinga, Dolan, & van der Molen, Reference Huizinga, Dolan and van der Molen2006), peak performance is achieved approximately between the ages of 21 and 30 (Cepeda, Kramer, & Gonzalez de Sather, Reference Cepeda, Kramer and Gonzalez de Sather2001), during which the frontal-striatal networks maturation contributes to both cognitive and behavioral flexibility processes (Morris et al., Reference Morris, Kundu, Dowell, Mechelmans, Favre, Irvine and Voon2016). The maturation of EF during adolescence may have interesting associations with higher levels of risk-taking behavior observed during this stage, as teenagers shift their priorities to avoid peer rejection, and inhibit health or legal concerns to engage in more risky behaviors (Blakemore, Reference Blakemore2018). Studies employing a latent-factor approach suggest that a single unified EF factor may best describe the capacities of young children up to ~8 years of age, however by age 10, two separable EF components (putatively memory and ‘general’ EF) may be identified using statistical models (Brydges, Fox, Reid, & Anderson, Reference Brydges, Fox, Reid and Anderson2014). In young adults, a model of three correlated factors, i.e., working memory, inhibitory control, and CF, emerges as the best EF model, termed ‘unity and diversity’ (Friedman & Miyake, Reference Friedman and Miyake2017). Although this data fits the narrative of a developmental transition from relatively undifferentiated unidimensional EF to separable but correlated EF components, the relative paucity (and specificity) of age-appropriate tasks to measure emerging EF during infancy and childhood confounds this interpretation. Many child EF tasks require language skills to comprehend task instructions, which precludes their use in preverbal children and presents additional non-EF related task demands. There are mental health disorders which affect and impair the development of cognitive flexible thinking which start in childhood or adolescence, such as ASD (Hughes et al., Reference Hughes, Russell and Robbins1994), OCD (Gottwald et al., Reference Gottwald, de Wit, Apergis-Schoute, Morein-Zamir, Kaser, Cormack and Sahakian2018; Vaghi et al., Reference Vaghi, Vértes, Kitzbichler, Apergis-Schoute, van der Flier, Fineberg and Robbins2017), and schizophrenia (Murray et al., Reference Murray, Cheng, Clark, Barnett, Blackwell, Fletcher and Jones2008). Indeed, children and adolescents with OCD have impaired functioning at school and at home and experience severe distress. Critical cognitive domains for daily functioning and academic success are learning, memory, CF and goal-directed behavioral control. These domains, particularly learning and memory as well goal-directed control and cognitive plasticity are impaired early in the development of OCD (Gottwald et al., Reference Gottwald, de Wit, Apergis-Schoute, Morein-Zamir, Kaser, Cormack and Sahakian2018). In adults with OCD, a severe impairment in CF has been shown. Moreover, this impairment is likely due to disruptions in the fronto-striatal circuitry (Vaghi et al., Reference Vaghi, Vértes, Kitzbichler, Apergis-Schoute, van der Flier, Fineberg and Robbins2017) that typically subserve CF.

Older adulthood

As individuals age, cognitive abilities typically decline, including CF (Murman, Reference Murman2015). This CF decline presents as increased perseverative behaviors measurable in tests such as the Wisconsin Card Sort Task (WCST, Fig. 2c) (Ashendorf & McCaffrey, Reference Ashendorf and McCaffrey2008). A heightened propensity for older adults to perseverate can be linked to their declined set-shifting capabilities (Ridderinkhof, Span, & van der Molen, Reference Ridderinkhof, Span and van der Molen2002). At the neural level, the PFC experiences significantly greater gray matter volume loss during normal ageing as compared to other developmental stages (Raz et al., Reference Raz, Gunning, Head, Dupuis, McQuain, Briggs and Acker1997), which contributes to the age-related decline of EF and CF. Age-related neurodegenerative diseases, such as mild cognitive impairment, dementia, and Alzheimer disease (AD), can speed up neuronal dysfunction and exacerbate cognitive declines, including CF (Guarino, Forte, Giovannoli, & Casagrande, Reference Guarino, Forte, Giovannoli and Casagrande2020). Recent functional imaging studies investigating CF-related brain network dynamics suggest that older adults who have extended dwell time in co-activation among the lateral frontoparietal network (L-FPN, ‘executive control’ network) and medial frontoparietal network (M-FPN, ‘default’ network) show diminished CF, compared with young adults (Kupis et al., Reference Kupis, Goodman, Kornfeld, Hoang, Romero, Dirks and Uddin2021).

In older adults, CF also appears to have a significant impact on motor control. Older adults demonstrate reduced motor flexibility when switching between different walking patterns, which is positively associated with higher levels of cognitive perseveration (Sombric & Torres-Oviedo, Reference Sombric and Torres-Oviedo2021). This connection suggests a shared mechanism that governs both cognitive and motor perseveration as individuals age. As older adults with diminished CF (measured by WCST) have a greater risk of losing balance and falling (Pieruccini-Faria, Lord, Toson, Kemmler, & Schoene, Reference Pieruccini-Faria, Lord, Toson, Kemmler and Schoene2019), interventions focusing on enhancing cognitive and motor flexibility could be instrumental in maintaining the quality of life for older adults by preventing falls.

More longitudinal data are required to understand CF across the lifespan, however when examining young children, there are some data to suggest that EF assessed by a battery of tests at 24 months does relate to EF assessed at age 4 (see Miller, Galvagno, & Elgier, Reference Miller, Galvagno and Elgier2023). Stability prior to that is unclear as studies have given mixed results. There are cross sectional data on CF across the lifespan using the CANTAB IED, which shows that CF performance improves from childhood to adolescence and is optimal in early young adulthood and then remains stable until age about 50 and slowly declines during older adulthood (Langley et al., Reference Langley, Sahakian, Robbins, Boyle, Stein, Stern, Sahakian, Golden, Lee and Chen2023).

Quality of early caregiving

The quality of early caregiving is an important and modifiable factor that affects early CF development. Parental social interactive behaviors, including sensitivity, have been linked to higher CF abilities in children (Bernier, Carlson, Deschênes, & Matte-Gagné, Reference Bernier, Carlson, Deschênes and Matte-Gagné2012) whilst traumatic social experiences can negatively impact children's EF, including CF (Kavanaugh, Dupont-Frechette, Jerskey, & Holler, Reference Kavanaugh, Dupont-Frechette, Jerskey and Holler2017). One promising parent-based EF training program is the Attachment and Biobehavioral Catch-up intervention for infants (ABC-I). This 10-session home-based program fosters nurturing and synchronous parent-child interactions (Dozier & Bernard, Reference Dozier and Bernard2017) and has been shown to positively affect attachment security, emotion expression, and cognitive control in children whose foster parents underwent training (Bernard, Hostinar, & Dozier, Reference Bernard, Hostinar and Dozier2015). These encouraging data highlight the potential for home-based programs that target parent-child relationships to enhance early EF development.

Lifestyle factors

A range of lifestyle factors, such as sleep, exercise, nutrition, stress management, social connection, and learning new skills, can enhance EF and combat the cognitive decline associated with ageing (Beddington et al., Reference Beddington, Cooper, Field, Goswami, Huppert, Jenkins and Thomas2008). Here, we focus on stress-management via mindfulness and exercise as an illustrative example for potential intervention. Chronic and acute stress can negatively affect CF throughout one's life. For example, stress can hinder attentional shifting in infants (Seehagen, Schneider, Rudolph, Ernst, & Zmyj, Reference Seehagen, Schneider, Rudolph, Ernst and Zmyj2015) and task-switching abilities in young adults (Plessow, Kiesel, & Kirschbaum, Reference Plessow, Kiesel and Kirschbaum2012). Moreover, the brain is particularly vulnerable to toxic stress during early life and older adulthood (Lupien, McEwen, Gunnar, & Heim, Reference Lupien, McEwen, Gunnar and Heim2009), so preventative interventions are indicated during these life stages. Research suggests that mindfulness practice and exercise may be efficacious in reducing stress (Vatansever, Wang, & Sahakian, Reference Vatansever, Wang and Sahakian2021) and boosting EF (Lerche et al., Reference Lerche, Gutfreund, Brockmann, Hobert, Wurster, Sünkel and Berg2018). Notably, exercises incorporating mindfulness, like Tai-Chi and Taekwondo, are more effective in improving EFs compared to standard resistance and aerobic exercises (Diamond & Ling, Reference Diamond and Ling2020). One explanation is that mindful exercises demand greater cognitive control, but further research is needed the clarify mechanistic relationships between CF, mindfulness, and general well-being.

Language factors

The relationship between bi-/multilingualism and EFs, including CF, has generated significant scientific discussion. Studies suggest that bilinguals maintain constant activations for both their languages (Thierry & Wu, Reference Thierry and Wu2007) and these continuously active language representations compete for selection during language use, necessitating monitoring and control from bilinguals to achieve successful communication (Valian, Reference Valian2015). Mechanisms of selection, inhibition and shifting operate in tandem to manage interference but also facilitate language switching when necessary (Gallo, Novitskiy, Myachykov, & Shtyrov, Reference Gallo, Novitskiy, Myachykov and Shtyrov2021). Clearly, bilinguals need to exercise extensive linguistic control to use their languages effectively, which may be related to domain-general cognitive control, as measured in EF tasks. Despite active research, results remain inconclusive. Several behavioral and neuroimaging studies report better EF performance in bilinguals (Barac, Moreno, & Bialystok, Reference Barac, Moreno and Bialystok2016), but others dispute the existence of any advantage (Duñabeitia et al., Reference Duñabeitia, Hernández, Antón, Macizo, Estévez, Fuentes and Carreiras2014). This lack of consensus may be due in part to inconsistent measurement of bilingualism as a binary variable rather than as a multifactorial continuum, which obscures deeper differences between monolingual and bilingual groups (Kaushanskaya & Prior, Reference Kaushanskaya and Prior2015). To advance the field, longitudinal studies and training studies that allow for more careful dissection of causal relationships between multilingualism and mental flexibility are needed. Indicatively, short second-language (L2) learning interventions yield benefits for attention switching in both children (Janus, Lee, Moreno, & Bialystok, Reference Janus, Lee, Moreno and Bialystok2016) and adults (Bak, Long, Vega-Mendoza, & Sorace, Reference Bak, Long, Vega-Mendoza and Sorace2016). Explicit training in a language-switching paradigm has shown potential transfer outside the linguistic domain, manifesting in reduced switching (Timmer, Calabria, & Costa, Reference Timmer, Calabria and Costa2019) and mixing costs (Liu et al., Reference Liu, Yang, Jiao, Schwieter, Sun and Wang2019) in adult bilinguals. Therefore, learning a new language may potentially be an effective strategy for improving CF throughout the lifespan.