Can Interacting with Animals Improve Executive Functions? A Systematic Review

Simple Summary Executive functions are cognitive processing skills associated with planning, problem-solving, decision making, and regulating behaviour. For some individuals these abilities may be impaired, which can have negative long-term impacts. It has been proposed that interacting with animals may provide an opportunity to strengthen these skills. A systematic review was carried out to assess the ways in which interacting with animals may improve executive functions. This review included 23 studies exploring executive functions across three contexts: the human–pet relationship, the presence of an animal, and involvement in an animal-assisted service. There is some evidence to suggest that interacting with an animal may be beneficial for older adults, whilst horseback riding seems particularly beneficial for children; however, the overall methodological rigour is limited. Abstract There has been growing interest in the potential benefits of using human–animal interactions to improve executive functions: cognitive processes that allow individuals to plan, solve problems, and self-regulate behaviour. To date, no comprehensive review has been conducted. The purpose of this study was to evaluate existing literature, adopting broad inclusion criteria. Following Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines, 16 papers were identified from peer-reviewed literature. Additional papers were identified from grey literature, including 6 dissertations and 1 thesis. A review of these 23 studies found that human–animal interactions and executive functions are investigated in three main ways: (1) exploring the potential benefits of the human–pet relationship, (2) exploring the impact of an animal’s presence during administration of executive function tests, and (3) evaluating the efficacy of animal-assisted services (e.g., animal-assisted therapy) on executive functions. Five of the included studies reported a significant improvement across all measured domains of executive functions, but effect sizes were underreported. Comparatively, 9 studies reported mixed findings, d = 0.32–0.55, while 8 studies reported no significant results. The overall rigour of the research was limited, with great heterogeneity between the study methodologies and outcome measures used. It is recommended that future studies utilise high-quality research methodologies through the use of randomisation, pre- and postmeasures, and appropriate control conditions, where possible.


Introduction
'Executive functions' (EFs) is an umbrella term for a set of cognitive processes that allow individuals to regulate their behaviour, plan, solve problems, and make decisions [1,2]. While the conceptualisation of EFs is subject to ongoing debate [3], many developmental and cognitive researchers have defined EFs as comprising three core processes: (1) working memory (WM), the ability to monitor new information and revise old information; (2) inhi-

Materials and Methods
A literature search was conducted using the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines [45]. Relevant literature was first identified through electronic searches of databases, including Google Scholar, PubMed, Web of Science, Science Direct, and PsycINFO from inception to September 2022. The HABRI Central (habricentral.org/resources) and WALTHAM (waltham.com/resources) websites, which compile research on human-animal relationships, were also searched. The search terms used to collate the literature were 'animal-assisted intervention', 'animal-assisted therapy', 'pet', 'companion animal', and 'human-animal interaction', in combination with 'executive function *', 'working memory', 'inhibit *', 'cognitive flexibility', and 'attention'.
The following criteria were used to select relevant papers for review: (1) publication in English; (2) collection of empirical, quantitative data on overall EF ability, a specific EF process (e.g., working memory), or the prefrontal cortex with explicit reference to EFs; and (3) reference to human-animal interactions (HAI) or involvement in an animal-assisted service (AAS), such as animal-assisted therapy, animal-assisted education, and animalassisted activities. The participant demographic was kept broad across age, ability, and potential diagnosis, and different types of human-animal interactions (e.g., involvement in a therapeutic program or interaction with a pet) were explored. Nonpublished works were retained to avoid publication bias.
Studies were excluded if they did not include at least one adequate measurable assessment of executive function. For example, the Mini-Mental State Examination (MMSE) [46] is a screening tool for cognitive impairment in older adults and contains an item on working memory. However, the MMSE also assesses other cognitive domains and presents cognitive ability as a total score. As such, studies using the MMSE or other similar tools cannot be used to determine any impact on EF ability alone. Similarly, studies that focused on overall ASD, ADHD, or dementia symptoms were excluded if they did not specifically focus on EF.
The quality of the included studies was assessed, where possible, using various checklists designed by the Joanna Briggs Institute (JBI), including the Critical Appraisal Tool for Assessment of Risk of Bias for Randomised Controlled Trials [47], the Checklist for Quasi-Experimental Studies [48], and the Checklist for Case Reports [49]. By asking 8 to 13 closed-ended questions, the JBI checklists evaluate the quality of the study design, the quality of comparison groups, and the reliability of the outcome measures and statistical analyses. Appropriate JBI checklists were not available for all research designs included in this review, particularly studies in which the effects of the human-pet relationship were evaluated.

Study Selection
The initial search identified 41,866 articles, with Figure 1 showing the study selection process. Due to the inclusion of the search term 'attention', a large number of papers were collated on ADHD literature as well as literature exploring attention and regulatory behaviours in educational settings. This is not uncommon in EF literature, with other reviews reporting large initial searches [50,51]. As per our exclusion criteria, papers were excluded if they did not specifically focus on EFs.
The initial searching and screening were performed by the first author (DT) following a three-stage approach (title, abstract, full text), during which the sample was reduced to 476 potentially relevant articles. This sample included book chapters and literature reviews that were screened for additional records before being considered for inclusion; none were identified. Following the approach of other reviews with a larger search size, a randomised subset of the full-text papers (n = 73; 15%) were independently screened by the second author (JS). At the conclusion of the screening, a total of 16 peer-reviewed journal articles were retained, along with 6 dissertations and 1 thesis. The final sample size was therefore 23 works, published between 2014 and 2022. Of these studies, 16 (69.6%) were published after 2018. At time of publication, the corresponding authors were located in the United States of America (n = 8, 34.8%), the United Kingdom (n = 4, 17.4%), Israel (n = 3, 13.0%), Australia (n = 3, 13.0%), and Belgium, Italy, Japan, South Korea, and Switzerland (n = 1 each, 4.3%). a randomised subset of the full-text papers (n = 73; 15%) were independently screened by the second author (JS). At the conclusion of the screening, a total of 16 peer-reviewed journal articles were retained, along with 6 dissertations and 1 thesis. The final sample size was therefore 23 works, published between 2014 and 2022. Of these studies, 16 (69.6%) were published after 2018. At time of publication, the corresponding authors were located in the United States of America (n = 8, 34.8%), the United Kingdom (n = 4, 17.4%), Israel (n = 3, 13.0%), Australia (n = 3, 13.0%), and Belgium, Italy, Japan, South Korea, and Switzerland (n = 1 each, 4.3%).

Study Characteristics
A review of all included studies (N = 23) found that researchers explored EFs in three main ways: (1) examining associations between the human-pet relationship and EFs, (2) testing whether EF task performance is improved in the presence of an animal, and (3) examining EFs following involvement in an animal-assisted service. Due to these different research questions and therefore the differences in study design and findings, a direct comparison of these studies is difficult. It was also impossible to conduct a meta-analysis. Instead, the three types of research questions are discussed separately in the following analysis. Finally, 11 of the included papers (47.8%) did not provide effect sizes; where possible, when adequate data was presented, effect sizes were manually calculated [52,53].

Research Question 1: The Human-Pet Relationship
Four studies examined the human-pet relationship. Two studies examined the human-pet relationship in older adults, with sample sizes ranging from 52 to 88 participants and with approximately half of the participants from each study caregiving a pet. Participants were recruited from individuals receiving support for physical and/or mobility limitations [39] and stroke rehabilitation [54]. Two studies examined the human-pet relationship in children, with one of these testing whether children who engage in a greater number of household chores have stronger EF skills, with household chores including pet care-related activities (e.g., feeding a pet, taking a pet for a walk) [38]. The final study examined longitudinal data from the Avon Longitudinal Study of Parents and Children (ALSPAC) birth cohort study [55]. While the ALSPAC study initially recruited 14,541 pregnant women, only 13,557 participants provided initial data on whether they kept a pet. Of these participants, 58% reported keeping a pet during gestation. By the time the child was age 10, 74% of participants had a pet. This remained relatively stable, with 72% of the now adolescents (52% male) self-reporting living with a pet between the ages of 11 and 18 years [55].
The results suggest that the human-pet relationship may benefit older adults, with both Branson et al. [39] and Demeter [54] finding that older adults with a pet had stronger EF ability across executive control and sustained attention than older adults without a pet. In comparison, it does not appear that caring for a pet influences children's EFs. In the study by Tepper and colleagues [38], there was no evidence to suggest that engagement in pet care chores predicted EFs, whereas engagement in other household chore types (e.g., children making their own beds) was a predictor of EF ability. Supporting this, the longitudinal study by Purewal [55] found no significant developmental differences between individuals with a pet, regardless of the pet species, versus individuals without a pet. The results for the four studies exploring the human-pet relationship are shown in Table 1. Note. a = dissertation or thesis; b = study by Purewal [55] reports on participants recruited from longitudinal ALSPAC birth cohort study; c = broader ALSPAC research reports diagnoses including but not limited to ASD and ADHD [56,57] The quality of the four human-pet relationship studies could not be assessed using JBI checklists, which are not designed for longitudinal studies or cross-sectional research not pertaining to epidemiology and disease prevalence [49,58]. For these studies, we instead note the limitations inherent in these studies, including the use of convenience sampling [39,54] and that the cross-sectional data means that casual interpretations cannot be made [38,39,54,55].

Research Question 2: Presence of an Animal
Several studies (n = 6) examined whether interacting with an animal impacts performance on an EF task. These studies were further divided into two categories: completing EF tasks whilst in the presence of an animal (n = 4, 66.7%) [59][60][61][62] and completing EF tasks immediately after interacting with an animal (n = 2, 33.3%) [63,64]. Two of the studies did not provide enough detailed information to determine the time spent in the overall experiment (e.g., greeting researchers, receiving instructions, testing with animal present, finishing experiment) versus actual time spent directly interacting with the animal. Of the four studies that provided information about the total time spent with the animal [59,[62][63][64], interactions ranged between 3 min and 15 min in length (M time = 6.25, SD = 5.85).
For the three studies examining children, the mean age was 10.18 years (SD = 2.33), with two of these studies recruiting from populations with a diagnosis of neurodevelopmental and/or behavioural disorders [59] and a learning disability [61]. There appears to be some evidence that the presence of an animal can improve cognitive flexibility and inhibitory skills [59] and WM [61,63] for children. In addition, Hediger and Turner [63] found that brain activity, as measured by neurofeedback in the frontal lobe and the prefrontal cortex, the area of the brain associated with EFs, was greater when interacting with a real dog versus a robotic control.
For the studies exploring EFs in adults, all participants were recruited from university students (M age = 20.09, SD = 1.00). In a neuroimaging study by Nagasawa et al. [62], participants demonstrated greater prefrontal cortex activation when interacting with a cat, with the authors concluding that this may transfer to overall EF ability. In comparison, Thayer and Stevens [64] found no significant changes in EFs when interacting with a dog across two independent experiments, as measured by performance on WM tasks. Similarly, results from Gee et al.'s [60] counterbalanced study were mixed, with participants improving in EF task performance when either the human collaborator or the dog were simply sitting next to the participant. However, performance was poorest when participants were required to maintain physical contact with the dog compared to when they were maintaining physical contact with the human collaborator [60]. The study characteristics and results of all six studies are presented in Table 2.
As shown in Table 3, the overall quality of assessed studies was moderate to high, with all studies measuring outcomes and conducting analyses appropriately. However, for the four studies using a randomised controlled trial design, all were ambiguous in terms of how participants were randomly assigned to conditions. Due to the nature of the studies and the research question, blinding the participants to the presence of the animal was not possible; one study, however, did attempt to obfuscate the aim of the study by telling the participants that the experimenter needed to bring their dog to the testing session, as the dog could not be left alone at home [63]. Additionally, outcome assessors were not blinded to the conditions. All but one study used validated measures. In Gee et al.'s study [60], participants played a digital version of the classic electronic Simon game, wherein participants reproduce, in order, a pattern of lights presented on a four-quadrant touchpad; the Simon game has been well represented in past research, but this digital version has not been validated [65,66]. The two quasi-experimental studies, as seen in Table 3, were also moderate in quality, lacking the inclusion of appropriate control conditions.  Note. ITT = Intention-to-treat Analysis.

Research Question 3: Animal-Assisted Service
Thirteen studies examined the efficacy of animal-assisted services on EFs. Most of these studies examined the use of animal-assisted services with children (n = 10, 76.9%; M age = 9.51, SD = 9.04), while two (15.4%) studies used adult participants; of these two papers, one recruited from younger adults [67], while another paper recruited older adults [68]. Finally, one study examined the impact of an animal-assisted service on both adult and child participants [69]. Among all the studies, the sample size ranged from 2 to 309 participants (M = 66.85, SD = 86.43). For the 12 studies that provided information on the total program duration, the programs ranged from 4 to 52 weeks in length, with an average of Most of the reviewed studies involved horses and took place at a riding centre (n = 8, 61.5%). Excluding Norwood et al. [70] and Schroeder [69], 6 of the studies were designed to address specific therapeutic goals, such as improving social and communication skills, and self-regulation skills. Four studies (30.8%) incorporated a dog, with settings including schools (n = 2, 50.0%), a university campus (n = 1, 25.0%), and a clinic (n = 1, 25.0%). The number of dogs involved in these studies ranged from 2 to 27. Typically, the dogs were recruited from therapy dog organisations and accompanied by a trained handler. However, in 1 study the dogs were recruited from a nonprofit rescue organisation, with interactions overseen by a veterinarian [71]. Across all studies, interactions with the dogs varied. Finally, 1 study explored the benefits of raising 4-5 garden crickets over an 8-week period [68]. A research assistant ensured compliancy with the program through weekly telephone counselling [68]. The study characteristics and findings for all 13 studies are presented in Table 4.
The overall quality of the assessed studies was moderate to high. Overall, strengths of the studies included the use of reliable and validated measures of EFs and the use of appropriate statistical analyses. For the randomised controlled trials, there was once again some ambiguity over how participants were allocated to conditions, with only one study specifying that a computerised, random number generator had been used [68]. Due to the nature of the programs, blinding the participants and those delivering the programs to the participants was not feasible, although in one study the authors informed all participants that they would have the opportunity to interact with animals but blinded the participants as to the timing and amount of human-dog interaction they would receive [67]. This was done in an attempt to prevent condition-specific attrition [67]. Finally, related to blinding, it was often unclear whether the outcome assessors were aware of which conditions the participants had been assigned.
The quasi-experimental research and case studies were of moderate quality. A common limitation was the absence of a control condition. Norwood et al. [70] attempted to address this by testing a smaller subset of the participants six weeks prior to the program commencing, allowing participants to act as their own control across three time points (T0-T1-T2). The study by Schroeder [69] was the only study to include a traditional control condition; however, some of the participants in the experimental condition were already participating in horseback riding. Finally, the 2 case studies would have benefitted from more information about the participants [72] and the horseback riding program [73], respectively. Table 5 presents the quality assessments for the 13 studies examining the impact of animal-assisted services.    Note. a = dissertation or thesis; b = diagnoses included but not limited to ASD, ADHD, Down's syndrome, and global development delay; c = children attending alternative school due to previous suspension, expulsion, or behavioural difficulties; d = diagnoses included but not limited to ASD, ADHD, disruptive mood regulation disorder, and intellectual disability; e = work by Schroeder (2015)

Discussion
The purpose of this review was to explore whether interactions with animals can improve executive functions. A systematic review of peer-reviewed and grey literature yielded 23 studies addressing three broad research questions: (1) examining the association between the human-pet relationship and EFs, (2) testing whether EF task performance is improved in the presence of an animal, and (3) examining whether EFs improved following involvement in an animal-assisted service. Due to these different research questions as well as heterogeneity between research design, participants, outcome measures, and quality, a critical comparison of the literature was difficult. However, some trends did emerge.
First, three of the studies recruited older (60+ years) adults, with all studies suggesting that interacting with an animal can improve EFs across multiple domains, such as sustained attention and overall executive control [39,54,68]. In Branson et al.'s study, the directionality of the relationship between caring for a pet and EFs could not be determined; it is possible that caring for a pet improves EF skills, but it is also probable that older adults with better EFs are more likely to keep a pet [39]. Demeter's study provided some further evidence that caring for a pet can help retain EF skills in older adults, with significant differences found between stroke survivors living with a pet versus stroke survivors without a pet, but this study was limited by a small sample size and convenience sampling [54]. The strongest evidence for the benefits of caring for a pet for EFs comes from Park et al. [68], in which older Korean adults who raised pet crickets over an eight-week period demonstrated improved EFs compared to participants in the control condition. Taken together, these three studies provide evidence supporting the benefits of the human-pet relationship on EFs for older adults. Due to the paucity of this research, however, further studies are clearly needed.
There was no research exploring the benefits of caring for a pet for adults aged 18-59 years, although two studies in this age group explored whether the direct presence of an animal could improve performance on WM tasks [60,64], and two additional studies explored the benefits of animal-assisted services [67,69]. Across these research questions, results were mixed. Nagasawa et al. found that interacting with a cat activated the prefrontal cortex, the brain area implicated in EFs, and postulated that this would generalise to improved EF skills [62]. In comparison, Thayer and Stevens [64] found no significant differences in WM skills when completing the task in the presence of an animal, while Gee et al. suggested that WM performance was poorest when participants were required to maintain contact with a dog [60]. For the long-term intervention studies, only Pendry and colleagues found significant improvements in EFs across time [67]. In addition to reflecting differences across study duration and time spent with an animal, this may also reflect that Pendry et al. recruited at-risk adult participants, including those with learning disabilities or mental health conditions, who may have benefitted more from EF training [67].
Two studies exploring the benefits of caring for a pet for children did not find any significant differences between pet caregivers and non-pet caregivers [38,55], unlike the research with older adults. One possible explanation for this difference is that adults are more likely to be responsible for pet care, with activities such as remembering to provide the correct amount of food for a pet possibly providing an opportunity to practise EF skills [38,39]. In comparison, young children are less likely to be the primary caregiver for a pet, while some research has suggested that adolescents become less involved with their pets as they get older [80,81]. While caring for a pet did not appear to improve EFs in children, some of the reviewed research suggests that the simple presence of an animal can improve task performance [59,61,63]. It is possible that the presence of an animal may reduce children's feelings of stress and anxiety and provide a nonjudgemental source of support, thereby improving task performance, with this explanation arising from similar research exploring the benefits of reading-to-dog programs [41].
For research examining improvements to children's EFs across time, several studies suggested that horseback riding improved EFs, including inhibition, WM, monitoring, and planning skills [70,74,77,78]. This supports previous research that found that horse-back riding and equine-assisted therapy can improve self-regulatory behaviours, with a previous review suggesting improvements in hyperactivity, irritability, and task engagement in autistic children and adolescents [82]. The research on therapeutic programs incorporating a dog was less consistent; in one study, participants with a poorer baseline demonstrated significant improvements in EFs [79], while two studies found no significant differences [71,76]. Interestingly, most of the studies that reported improvements in EFs were longer-term programs, taking place over more than seven weeks [70,[73][74][75]77,78], while a six-week program [72] and a four-week program [76] reported mixed and no significant findings, respectively.
Different research designs and outcome measures make it difficult to determine whom these interventions most benefit. It is also noteworthy that two studies, which measured EFs using parent reports and reports completed by horseback riding instructors, found that parents reported no significant improvements in their child's EFs [69,78]. This may reflect that some EF skills do not transfer to other contexts, such as the home, or reflect bias within the instrument. In addition, the longer-term benefits of animal-assisted services remain unknown, with few of the reviewed studies conducting follow-up. A review by Diamond and Lee [21] suggested that the impacts of EF interventions diminish over time once the intervention is ceased, and it is possible that interacting with an animal may have no long-term benefits on the development of cognitive skills.
Overall, the studies exploring the potential benefits of human-animal interactions on executive functions was of moderate to high quality. We highlight that future research should address potential bias within measures and explore longer-term changes in EFs, while the broader human-animal interaction field would benefit from the inclusion of larger sample sizes, the use of randomised controlled trials, and appropriate randomisation and control conditions [83]. Unfortunately, we were unable to provide effect sizes for several papers in this review because the effect size had not been explicitly reported or the original paper did not present the appropriate statistics to compute an effect size. This has previously been highlighted as a concern in other human-animal interaction reviews [41,84], as statistical significance alone is not enough to determine the efficacy of using human-animal interactions to improve health, well-being, and functioning.
There are several limitations inherent in the current review. For example, the conceptualisation of EFs was limited to three core cognitive constructs: working memory, inhibition, and cognitive flexibility. While this three-factor model of EFs is prevalent within the literature [6], some researchers have suggested that EFs encompass additional constructs or domains [85]. In the present study, the inclusion of 'executive function' as a broad search team as well as our large initial search sample likely captured all papers exploring the relationship between human-animal interactions and EFs, but it is possible that some papers using a different terminology were missed. In the present study, we also only reviewed quantitative literature, thereby possibly missing research examining related constructs such as on-/off-task behaviour and regulatory behaviours [86]. Future reviews may therefore benefit from expanding the search terminology and exploring qualitative and observational research.

Conclusions
As executive functions are linked to academic and workplace success, adaptive functioning, and quality of life, it is important to continue exploring ways to improve these skills. Overall, the effects of interacting with animals on executive functions cannot be clearly determined from the current literature; however, there are some promising trends. In particular, it appeared that interacting with animals, encompassing caring for a pet and animal-assisted services, may benefit older adults. Additionally, there was some evidence to support the efficacy of horseback riding programs, particularly for at-risk children and adolescents. Higher-quality studies exploring the benefits of the human-pet relationship and the benefits of involvement in animal-assisted service are recommended, with a focus on incorporating randomised controlled trials or longitudinal study designs.