Mindfulness in Face Recognition: Embedding mindfulness instructions in the face-composite construction process

Summary Meditative mindfulness practices, promoting sustained attention and reducing mind-wandering, have been associated with improvements in cognitive abilities and memory. The present study explored whether a non-meditative practice could be successfully applied in a forensic application; specifically, whether mindfulness instructions can be embedded in the face-composite construction process to facilitate identification. Twenty participants, who were not football fans, were asked to memorise an unfamiliar footballer's face and return 24 hr later to construct a face using the self-administered EvoFIT facial composite system. In the experimental condition, mindfulness instructions were embedded in the EvoFIT system, encouraging witnesses to focus on the target face and the process; in the control condition, participants constructed the face using the standard EvoFIT system. Naming of the composites was attempted by 24 football fans, who each viewed 10 composites, five from each condition, and then the target footballer images to ensure they were familiar with the identities. Results showed significantly higher levels of correct naming for composites constructed using EvoFIT with mindfulness instructions compared to using the standard EvoFIT. These findings indicate the potential for non-meditative mindfulness instructions to assist face-composite construction, improving correct naming of ensuing composites.


Summary
Meditative mindfulness practices, promoting sustained attention and reducing mind-wandering, have been associated with improvements in cognitive abilities and memory. The present study explored whether a non-meditative practice could be successfully applied in a forensic application; specifically, whether mindfulness instructions can be embedded in the face-composite construction process to facilitate identification. Twenty participants, who were not football fans, were asked to memorise an unfamiliar footballer's face and return 24 hr later to construct a face using the self-administered EvoFIT facial composite system. In the experimental condition, mindfulness instructions were embedded in the EvoFIT system, encouraging witnesses to focus on the target face and the process; in the control condition, participants constructed the face using the standard EvoFIT system. Naming of the composites was attempted by 24 football fans, who each viewed 10 composites, five from each condition, and then the target footballer images to ensure they were familiar with the identities. Results showed significantly higher levels of correct naming for composites constructed using EvoFIT with mindfulness instructions compared to using the standard EvoFIT. These findings indicate the potential for non-meditative mindfulness instructions to assist face-composite construction, improving correct naming of ensuing composites.

K E Y W O R D S
EvoFIT, face recognition, face-composite construction, mindfulness

| INTRODUCTION
An eyewitness of a crime may be asked by police to assist in creating an impression of the face of a perpetrator, an image commonly known as a facial composite (Davies & Valentine, 2007). A facial composite is typically constructed with the help of a trained forensic practitioner (Frowd, 2015) and/or computerised systems such as EvoFIT (Frowd et al., 2004), EFIT-V/6 (e.g., Davis et al., 2016) and ID (Tredoux et al., 2006). As opposed to feature-based systems (e.g., E-FIT and PRO-fit), where the process entails combining face features (i.e., eyes, nose, mouth, etc.), EvoFIT, EFIT-V/6 and ID follow a holistic approach that encourages processing the face as a whole that is more effective in face-composite construction (e.g., Wells & Hasel, 2007). While these holistic systems have all been evaluated in the literature (e.g., Davis et al., 2016;Frowd et al., 2019;Tredoux et al., 2006), EvoFIT is notably accompanied by research exploring additional methods, such as mindfulness meditation (Martin et al., 2017), to assist with composite construction and, in effect, composite identification. The present research focused on enhancing the EvoFIT process further by embedding non-mediative mindfulness instructions throughout the face-composite process, to assist with face construction.
EvoFIT has been successfully integrated into forensic processes, leading to the successful conviction of offenders (e.g., Frowd, Pitchford, et al., 2011;Frowd et al., 2019). The 'gold-standard' procedure involves a witness participant viewing a photograph (or video) of an unfamiliar person, and, usually from 24 to 48 hr later, working to create a composite of the face the witness remembers. The face construction process begins with witnesses repeatedly selecting faces from arrays of faces, which are combined to 'evolve' a face. Then, software tools are used to manipulate the face holistically (e.g., weight or age) and alter the size and position of facial features. The task concludes by adding external features (e.g., hair). The final composite is shown to the public (or other participants who know the person depicted), to attempt recognition (Fodarella et al., 2015).
Research has investigated methods to enhance the EvoFIT composite process, focusing on improving recollection of the target face. The Cognitive Interview (CI), employed at the beginning of an EvoFIT session, invites the witness to recall as much detail about the crime scene and the offender as possible (Fisher & Geiselman, 1992 by visualising the target face and freely recalling as many details of him or her as possible (Fodarella et al., 2015;Frowd, 2011). The Holistic Interview (HI), employed after the CI, asks the witness to make judgements about the target based on specific personality traits (e.g., intelligence, aggressiveness; Frowd et al., 2008). Combined with EvoFIT, both interviewing procedures (together known as H-CI) are valuable for identifying suspects (e.g., Frowd et al., 2012). Recently, Martin et al. (2017) incorporated a brief focused breathing exercise before applying CI and H-CI techniques.
Focused breathing inductions guide participants to direct their attention and awareness to present moment sensations, focusing on the experience of breathing. When their awareness moves away from the breath, participants are guided back to the sensations of breathing (e.g., Arch & Craske, 2006). Focused breathing practices have been related to a reduction in stress (e.g., Chiesa & Serretti, 2009), a side-effect of experiencing a criminal event (e.g., Morgan et al., 2004), and constructing a facial composite . Martin et al. showed correct naming of facial composites created after the focused breathing exercise was significantly higher than of facial composites created using the standard procedure conducted without focused breathing. Focused breathing is an example of traditional mindfulness meditation training. Mindfulness meditation involves intentionally observing the breath (or any other attentional focal point), without reacting to feelings and thoughts that may arise, by accepting things as they are (Kabat-Zinn, 1990). Mindfulness training involves the recurrent practice of self-regulating attention, leading to measurable increases in attentional skills in non-meditating participants (e.g., Campillo et al., 2018;Chambers, Lo, & Allen, 2008;Semple, 2010;Tang et al., 2007). Meditative practices often require weeks of training (e.g., Geiger et al., 2016;Hilton et al., 2017), but research has also developed brief exercises, which can be effective in enhancing state levels of mindfulness (e.g., Carmody & Baer, 2009).
Mindfulness meditation has been associated with improvements in cognitive abilities (Chiesa et al., 2011) and memory (Levi & Rosenstreich, 2019), all mainly due to fostering of sustained attention and reduction in mind-wandering (e.g., Mrazek et al., 2012;Tang et al., 2007). Specific to brief meditative practices, mindfulness training has been linked to improved long delay free recall (but not long delay cued recall), better object recognition memory, and recognition memory in a word task (Brown et al., 2016;Rosenstreich & Ruderman, 2016), and better source monitoring of misinformation (Alberts et al., 2017) and memory recall (Hammond et al., 2006) in the eyewitness process. Moreover, a single mindfulness meditative session has been shown to improve short-term memory for faces (Youngs et al., 2020), recall of novel words (Alberts & Thewissen, 2011;Bonamo et al., 2015), attention to and memory of visual and auditory stimuli (Campillo et al., 2018) and fewer false memories in word tasks (Calvillo et al., 2018;Lloyd et al., 2016;but see Rosenstreich, 2016). Nevertheless, participants may cease meditation practice, or even refuse to engage with it, due to preconceptions around meditation or because of potential conflicts with their cultural and religious background (e.g., Mantzios & Wilson, 2013;Wellings, 2015).
Non-meditative practices have been developed to induce a state of mindfulness, without the commitment or engagement required in meditation, and are typically embedded within the cognitive or behavioural aspect of interest. For example, as an alternative practice to meditation, mindfulness instructions have been employed in the process of washing hands and dishes (Gilmartin, 2016;Hanley et al., 2015) and in the process of adding colour on lined art (i.e., colouring; Mantzios & Giannou, 2018a), both to successfully enhance mindfulness and reduce anxiety. In targeting specific behaviours, mindfulness instructions during the process of trying on a bathing suit have been found to effectively induce both a state of mindfulness that abated the influence of body satisfaction on negative effect and smoking urges (Adams et al., 2013;Bowen & Marlatt, 2009) and have been incorporated in food diaries to promote more mindful eating habits, again increasing mindfulness and reducing anxiety (Hussein et al., 2017;Mantzios & Wilson, 2014. Hence, recent research has collectively explored the potential of alternative practices that may appeal to a wider part of the population, and successfully identified ways of inducing mindfulness through nonmeditative practices, necessary in assisting specific processes, which in the present research could be utilised to create a facial composite. Recently, a standalone (i.e., self-service) version of the original EvoFIT facial composite system was developed, to be easily deployed to witnesses on a computer within a few hours of witnessing. The importance of this approach is threefold: first, as the process lasts about an hour, composite construction with this version of EvoFIT is time efficient. As facial composites are more effective when face construction occurs within hours rather than days (Frowd et al., 2005), the 'self-administered' system may help to improve composites and, thus, offender identification. Second, efficiency also applies to reduced demands for police resources, removing the need for comprehensive training of police officers and forensic practitioners. A self-administered procedure could be useful in cases of less serious crimes, easing the strain on police resources. Third, visiting a police station can conceivably elevate stress levels (Risan et al., 2016), adversely affecting witness performance (Kieckhaefer et al., 2014). As the standalone EvoFIT does not require a trained forensic practitioner, who would apply CI and H-CI processes (or a practitioner who would apply a mindfulness exercise), further research appears necessary to explore practices that could potentially enhance the outcomes of the self-administered construction process.
The self-administered EvoFIT system incorporates user prompts throughout the initial process of selecting faces from face arrays to encourage witnesses to visualise the face, instigating recall of the target face. Martin et al. (2017) suggested face construction benefited from mindfulness meditation; importantly, enhancing these already existing prompts through mindfulness instructions could result in gaining the cognitive benefits of mindfulness without the need for a practitioner to lead a meditative session prior to composite construction. Mindfulness instructions can benefit face-composite construction in two ways. First, mindfulness prompts, which encourage witnesses to close their eyes when visualising the face, introduce eyeclosure (i.e., an instruction to close the eyes or an automatic impulse to close the eyes) to improve memory for events (e.g., Perfect et al., 2008;Vredeveldt et al., 2010;Vredeveldt & Penrod, 2013). As part of meditative practices, eye-closure has been observed to facilitate free recall of an event without an increase in errors (e.g., Wagstaff et al., 2004Wagstaff et al., , 2011 and researchers put forward the idea that its importance lies in reducing general cognitive load (e.g., eliminating any monitoring of the environment; Perfect et al., 2008;Sprawson et al., 2020), resisting distractions (e.g., Perfect et al., 2011) and facilitating visualisation (Vredeveldt et al., 2011). 1 Second, a task-congruent practice has the potential to cultivate mindfulness without the need to engage in meditation. For example, Mantzios and Wilson (2014) used diaries to successfully prime participants to develop more mindful eating habits, and Hanley et al. (2015) adapted dishwashing instructions to reproduce a mindfulness practice.
One element that amplified the outcomes for participants in these studies was how these mindfulness practices were task-congruent, targeting specific activities and behavioural outcomes by aligning the intervention to the task, rather than using a generic contemplative intervention, separate from a task. As opposed to a mindfulness meditative practice before the task, task-congruent mindfulness instructions recurrently prompt the mindful cycle of acknowledging a distraction as a distraction and bringing awareness back to the present experience (e.g., Kabat-Zinn, 2003). Such a mindfulness-inducing cycle has the potential to be more beneficial in providing the necessary skills to focus attention to the composite construction process (e.g., Campillo et al., 2018;Mrazek et al., 2012), by resisting distractions (e.g., Diaz, 2013) and encouraging recall of the target face (e.g., Alberts & Thewissen, 2011;Hammond et al., 2006), during the process when these qualities are relevant (e.g., Adams et al., 2013).
In the present experiment, participants, who were not football fans, were asked to memorise a famous footballer's face, an identity who was unfamiliar to them, and return 24 hr later to construct a face using the self-administered EvoFIT facial composite system. In the experimental condition, mindfulness instructions were embedded in the EvoFIT system, encouraging witnesses to focus on the target face and the process; in the control condition, participants constructed the face using the normal EvoFIT procedure. The resulting face composites were presented to football fans, who attempted to name the footballers depicted in the composites. These football fans were further asked to name the famous footballer face images (i.e., the actual face images that the composite construction participants were asked to memorise), to ensure that they were familiar with these identities. We expected that embedding mindfulness instructions in the EvoFIT process would promote more identifiable composites.

| Composite naming
To ensure naming participants were familiar with the target footballers, the naming sample consisted of participants who correctly identified at least eight footballers from the target images (presented to naming participants after composite naming). Starting with an opportunity sample of 52 volunteer football fans, aged 20-48 years (40 males; M = 29.58, SD = 7.56), the final sample consisted of 24 male participants, aged 20-47 years (M = 30.67, SD = 6.70), who correctly identified eight or more of the famous footballers. The size of both samples is similar to previous research in the field (see Frowd, Pitchford, et al., 2011;Martin et al., 2017).

| Target faces
The stimuli were images of 10 famous, white male English footballers showing a neutral facial expression, without any distinguishing characteristics such as football team identifiers or hairstyles. Two sets of these 10 colour photographs, approximately 10 cm high by 10 cm wide, were printed on A4 paper. All 20 photographs were labelled at the back with a different number (randomly assigning numbers and, thus, participants to either the mindfulness instructions or standard condition) and were placed into one large envelope, for each participant to randomly select a face to construct. Once a face was selected from the envelope, it was placed in a separate envelope to ensure the specific target image, randomly assigning participants in each condition, was not used again. Hence, two composites were created for each target; one composite was created by a participant in the mindfulness instructions condition and one composite was created by a participant in the standard condition.
Composites were constructed on a laptop computer using selfadministered EvoFIT (v1.6.70) software. At the end of the study, two sets of images were collated to be used in naming of the composites.
Each set comprised 20 images. First came 10 of the 20 composite images, which were collated by a third independent party; to ensure naming participants were exposed to composites made in both conditions, each image set included five composites constructed in the mindfulness instructions condition and five composites constructed in the control condition. Moreover, as each footballer image was used to create two composites-one in the mindfulness instructions condition and one in the control condition-each of these 10 composites sets included one of the composites made of each footballer identity. The purpose of this design was to ensure naming participants would consider each composite solely as a unique identity and not in comparison with the preceded composites. In each set, the composite images were followed by the 10 target footballer images.

| Personality measure
The five facet mindfulness questionnaire-Short form The five facet mindfulness questionnaire-short form (FFMQ-SF) (Bohlmeijer et al., 2011) is a 24-item questionnaire measuring five facets of mindfulness, based on the original 39-item FFMQ (Baer et al., 2006).
The five mindfulness facets are; observing (e.g., 'I pay attention to physical experiences, such as the wind in my hair or the sun on my face'); describing (e.g., 'When I feel something in my body, it's hard for me to find the right words to describe it'); acting with awareness (e.g., 'I rush through activities without being really attentive to them'); non-judging (e.g., 'I think some of my emotions are bad or inappropriate and I shouldn't feel them'); and non-reacting (e.g., 'When I have distressing thoughts or images, I just notice them and let them go'). All item responses range from 1 (never or rarely true) to 5 (very often or always true), with higher scores indicating higher levels of mindfulness. Bohlmeijer et al. (2011) reported good reliability for all facets; observing, α = .81; describing, α = .87; acting with awareness, α = .83; non-judging, α = .83; non-reacting, α = .75. In the present experiment, Cronbach's α for total FFMQ-SF was α = .83 and Cronbach's α for the five facets was: observing, α = .78; describing, α = .71; acting with awareness, α = .62; non-judging, α = .73; non-reacting, α = .77.

| Procedure
We invited participants, who were not football fans, to attend two separate sessions over two consecutive days. On Day 1, we asked participants to first complete the FFMQ-SF as a baseline measure of their trait mindfulness levels, to ensure any observed effects were not driven by participants' already elevated trait mindfulness levels. Then, we asked them to randomly select a face from the envelope including target pictures. If the target face was unfamiliar to them, we asked them to study the face for 30 s. If the face was familiar to them, they were asked to choose another, encoding the first unfamiliar face (in the present experiment, one participant was familiar with their first choice, however, the next randomly selected face was unfamiliar and, thus, encoded). Their random selection of the face determined if they were placed in the experimental or the control condition group. After face encoding, participants were informed that they would attempt construction of the face the following day.
Participants returned 24 hr later to construct a composite of the face from memory using self-administered EvoFIT. On Day 2, we first briefly introduced the EvoFIT process to participants. All participants were encouraged to take their time and pay attention to all the instructions and subsequent prompts. The researcher remained available to answer any questions relating to the procedure. Fodarella et al. (2015) provide a detailed description of the standard EvoFIT face construction procedure, inclusive of CI and H-CI processes. In the present study, participants used a self-administered (standalone) EvoFIT protocol (e.g., Martin et al., 2017), which did not include the CI and H-CI procedures due to the standalone EvoFIT not requiring a forensic practitioner, who would apply these procedures.

| EvoFIT construction stage
Participants first viewed an instructional video briefly describing the process of creating a composite. In the standard condition, once the instructional video finished, participants were directed to the start of the face-composite construction process. In the standard condition, participants were presented with prompts throughout the process of face selection (prior to a screen of faces being shown), asking participants to simply visualise the face.
In the experimental condition, the video was followed by onscreen brief mindfulness instructions, discussing how the mind often wanders naturally and the process of returning the focus to the present moment and the process (see Mantzios & Giannou, 2018b). Participants were told to repeat this process of returning their focus to the process when distracted and when prompted to take a focused breath and visualise the face. The transcript of the mindfulness instructions added on EvoFIT was as follows: So, to construct a composite, you will be asked to select faces from the screen and make other choices to allow you to create the best likeness possible.
You will also be asked to visualise the face, basically to see it in your mind. Each time you are prompted to visualise the face, please close your eyes and direct your attention to the face you remember.
However, you might find that any noise heard inside or outside of the room distracts you, or that your mind wanders onto other things.
This happens to everyone and is of course absolutely fine. Our minds do that naturally.
What we ask is that, should such distractions arise, observe and label your thoughts as 'thoughts', or emotions as 'emotions', without judging or evaluating them any further. Simply say to yourself 'I just had a thought' and return your attention to the present and continue selecting faces from the screen, judging the best likeness, visualising the face, etc.
These instructions were an introduction to subsequently presented prompts, appearing throughout face selection (prior to a screen of faces being shown) and asking participants to direct their attention and visualise the face they remember. These prompts were: Please take a focused breath and visualise the face.
Remember that we all get distracted from time to time.
If your attention shifts away from visualising or constructing the face, acknowledge the distraction and bring your attention back to the present. Therefore, the mindfulness instructions were presented at the beginning of the process, with the mindfulness prompts, mimicking the existing standard prompts, being administered throughout the process of face selection but ceasing when the process progressed to holistic manipulation of the face and adding external characteristics (i. e., mirroring the appearance of the existing standard prompts).
Next, participants selected a specific pool of faces to work with, based on the age, race and gender of the face they remembered (e.g., white males, around 30 years of age). Based on the selected face database, randomly generated examples of faces (showing a face with only the internal characteristics-i.e., eyes, nose and mouth) were presented, asking participants to either proceed to face construction or return to the process of selecting a more suitable face database.
Once confident the database was a good representation of the target face, participants were presented with six successive arrays, each showing 18 faces (showing only the internal characteristics), asking them to choose two faces from each array, focusing on the eye region and ignoring face width. The first three arrays of faces presented smooth textured faces (i.e., faces without any shading and showing the same skin tone), whereas the subsequent three arrays presented textured faces (i.e., faces with differences in shading around facial features and skin tone). Once participants selected six smooth and six textured faces (i.e., two faces from each array), they were asked to pick one smooth and one textured face best matching the face they remembered.
Following, participants viewed two arrays of 18 faces, which were a combination (i.e., morphed together) of participants' selected faces, and were asked to select one face from each array. These two faces were, then, separately shown to participants and they had to select the best face of the two. The next page showed participants all selected faces and asked for a choice of the best matching face, which participants were asked to rate on a 1-to-10 likeness scale (1 being 'very poor likeness' to 10 being 'faces are identical'). The exact process of selecting two faces from arrays of 18 faces was repeated with 'evolved' smooth faces and 'evolved' textures of faces (i.e., where characteristics of the selected faces were 'bred together' to 'evolve' a face), again asking participants to select the best matching 12 faces (six for evolved smooth faces and six for evolved face textures), leading them to select the best combination face.
At this point, participants were asked to either evolve the face again, if they were not confident that the evolved face resembled the face they remembered, or, if confident, to continue the face construction manipulating the face through holistic tools. Participants used the holistic tools to alter the face holistically, across 14 categories (e.g., face weight, age, health, skin tone, extraversion, honesty, face position, etc.), by moving a slider left or right. Then, participants could use feature scales to adjust the position and shape of features (e.g., eyes, eyebrows, mouth, beard, etc.). Participants could use the holistic and feature scales as many times as required until they were confident the 'changed' face was a better match than the 'original' face. Finally, participants were asked to select hair (e.g., length, fringe, style, etc.) and other external features (e.g., hats and hoods) databases as many times as required in order to construct the best matching image of the target identity.
Once the participants were confident the final face with added external characteristics was a close match to the face they remembered, the researcher saved the composite face and exited the EvoFIT system.

| Naming stage procedure
The naming procedure lasted for approximately 20 min per person.
Participants were tested individually, where the researcher noted down responses. When, occasionally, two football fans attended a naming session together, each participant was randomly assigned a different set of composite and target images and was tested in isolation from the other participant, leading to individual responding to the naming task. Participants were informed that they would first see and attempt to name 10 composite faces of white English footballers who currently play or have in the past played internationally for the English national team. Next, participants were informed that they would see 10 famous footballer images to name, to ensure they were familiar with the targets. Composites and targets (see Figure 1 for examples of composites) were presented one at a time in a different random order for each person; no time limit was set for the naming process, with each composite and target remaining visible until a response was given (a correct name, explicit biographical information, or 'do not know'). Table 1 presents mean values of participants' baseline mindfulness levels (before face-composite construction) in the two conditions. An independent samples t test was conducted to compare mindfulness levels in the mindfulness and the standard conditions; there was no significant difference in mindfulness levels between the mindfulness and the standard conditions [FFMQ-SF; t(19) = .230, p = .820].

| Participant trait mindfulness levels
Mean values displayed for overall mindfulness score and subscales are comparable to those reported in previous research (e.g., Baer et al., 2006;Bohlmeijer et al., 2011;Moore et al., 2012).

| Composite identification
Responses on the composite naming task were scored as accurate identifications (i.e., when a composite or target picture had been correctly identified by a participant, either with the correct name of the footballer or with accurate biographical information), incorrect identifications (i.e., when a composite or target picture was incorrectly identified, providing an incorrect footballer name or incorrect biographical information) and 'Do not Know' responses (i.e., when no response was given). Table 2 presents descriptive statistics of these composite and target naming responses by Condition (Type of EvoFIT used to create the face).
In the analyses, 'conditional' scores were calculated for naming responses based on the number of composites and the number of targets that had been correctly named. This was done as participants would most likely not be able to correctly name a composite if that identity was unfamiliar to them. For example, if two composites and eight target images had been correctly named, the conditional naming score for this participant would be 2/8 or 0.25. Table 3  The Benjamini-Hochberg (1995) procedure was also applied, to minimise the possibility of Type I errors due to multiple comparisons (Abdi, 2010;Diz et al., 2011). The procedure offers a balance between false positives and false negatives, to increase the chances of finding true positives (Diz et al., 2011;Glickmann et al., 2014). Similar approaches have been applied in face recognition (Babaei et al., 2020) and mindfulness (Strohmaier, 2020;Strohmaier et al., 2020) research.
Correcting for three comparisons using this procedure, the aforementioned tests remained significant (or not).
By-items analyses were also carried out to explore naming levels for each item (composite) across participants (see Table 4). Pairedsample t tests revealed a marginally significant difference in favour of correct naming in the mindfulness over the standard condition, t (large effect size). Correcting for three comparisons using the Benjamini-Hochberg (1995) procedure, the aforementioned tests remained significant (or not).
As by-items analyses can be statistically weak, a unified byparticipants and by-items analysis was conducted on the correct naming responses using Generalised Estimating Equations. This regression-type approach emerged significant for type of EvoFIT [χ 2 (1) = 14.00, p < .001], 2 confirming that the mindfulness condition gave rise to more identifiable composites than those produced from the normal procedure [B = 1.47, SE(B) = 0.39, Exp(B) = 4.34, 95%CI (2.01, 9.37); indicating a large effect size].
Overall, these results suggest that incorporating mindfulness instructions into self-administered EvoFIT helps constructors to create more identifiable composites.

| DISCUSSION
The present study explored the effectiveness of embedding mindfulness instructions in the face-composite construction process, to promote more identifiable composites. Findings demonstrated significantly higher levels of correct naming for composites constructed using EvoFIT with mindfulness instructions compared to using the standard, self-administered EvoFIT system. Such a difference in performance was not observed with the number of mistaken Research has suggested eye-closure to facilitate free recall (e.g., Wagstaff et al., 2011) and improve memory for events (e.g., Vredeveldt & Penrod, 2013) and mindful attention and awareness skills to relate to improved long delay free recall (e.g., Brown et al., 2016), better memory recall in eyewitness processes (e.g., Hammond et al., 2006), and resisting distractions and mindwandering (e.g., Diaz, 2013;Mrazek et al., 2013). The present findings suggest that inducing these skills throughout the composite posites, compared to past research that utilised these procedures as part of the EvoFIT process (Frowd, 2011;Frowd et al., 2008). CI and H-CI processes are included to encourage witnesses to visualise the target face (Fisher & Geiselman, 2010) and make inferences about the character of the target face (Frowd et al., 2008), respectively.
The present mindfulness instructions and prompts invited witnesses to move away from being on automatic pilot, to visualise and focus on the face and the process of face construction, probing participants to attend to the present moment, non-judgmentally. These instructions and prompts echo some of the outcomes of the CI processes and focused breathing, which have been found to enhance face construction using the EvoFIT system (Frowd et al., 2008;Martin et al., 2017), enabling promotion and maintenance of the effects of CI and mindfulness throughout the construction process, to enhance accuracy in face construction. In effect, accurate naming in the mindfulness condition was at 15%, comparable to Martin et al.'s (2017) naming at 19%, considering they utilised the typical EvoFIT process, inclusive of CI and H-CI processes, but also the focused breathing exercise. However, inaccurate naming of the composites in the mindfulness condition was at 42%, higher than in the standard condition at 36%. Such difference, although non-significant, and considering that more 'Do not Know' responses were given for composites in the standard condition, possibly indicates that the composites constructed in the mindfulness condition were better, inducing identification more often.
The present findings further support the potential for mindfulness to be developed through non-meditative techniques, to support cognitive processes such as face recall and face-composite construction. Previously, Hanley et al. (2015) showed how an everyday activity such as dishwashing can become an informal mindfulness practice, and Mantzios and Giannou (2018a) (Giannou et al., 2020).
The present findings should be interpreted with caution due to the size of our composite naming sample. Although similar sample sizes have been reported in previously published work (e.g., Frowd, Pitchford, et al., 2011;Martin et al., 2017), future research should validate these effects further, by recruiting more naming participants but ensuring they are keen football fans. In the present study, naming participants, who did not meet our sampling criterion of correctly identifying more than eight target footballers, provided considerably more incorrect or 'Do not Know' responses; therefore, it appears that naming participants need to strictly be committed football fans in order to be able to correctly name the composites. Being a devoted football fan, regularly attending games, perhaps means more prior exposure to the footballers, enhancing familiarity (see Vallano et al., 2019). However, considering that a person attempting to name a composite might not be too familiar with a perpetrator, future research could investigate composite construction and identification of a face not overly familiar to both witness and naming participants.
Future research should also be conducted to offer insight into the potential mechanism by which mindfulness instructions resulted in better naming of composites. Research has related being attentive or mindful of an event to accurately recalling of information (e.g., Higgs & Donohoe, 2011), suggesting the potential of different attentional processes and practices to reinforce memory. Hence, future research could explore if the observed effects were a direct result of the intervention inducing the mindful cycle of acknowledging a distraction as a distraction and bringing awareness back to the present experience (e.g., Kabat-Zinn, 2003) or due to enhancing present moment attention to the process of face construction.
Mindfulness literature has exemplified two distinct fields within attention.
One field investigates what we would describe as sustained attention to the immediate present experience (e.g., Brown & Ryan, 2003). For Brown and Ryan (2003), mindfulness is openly experiencing the present moment that brings about present awareness and attention. The second field describes the sustained attention in the immediate present experience, but with the addition of a non-judgmental attitude towards thoughts and feelings that may arise in the present moment (Kabat-Zinn, 2003).
While both fields relate to models of attentional training and highlight the importance of maintaining attention in the present moment experience, the latter (applied to the present experiment) represents a model for the self-regulation of attention (Bishop et al., 2004) that results in overcoming mind-wandering and the consequences of thought suppression (paradoxically associated with an increased occurrence of thoughts; Abramowitz et al., 2001;Wegner, 1994) when non-judgement is present. The ability to adopt a non-judgmental, non-evaluative and accepting attitude enables the reiteration of attention to the present moment experience (e.g., Arch & Craske, 2006). Such a distinction between enhancing and sustaining attention and sustaining attention with the addition of self-regulating attention could be explored in the face-composite process, to identify the strongest approach to assist face composition and face recognition in general.
In conclusion, the present findings indicate the potential for nonmeditative mindfulness instructions to be embedded to the facecomposite construction process, without subjecting participants to further tasks or cognitive demands, and without the explicit instigation that witnesses are practicing mindfulness. The subjective differences between knowing and being blind to the such interventions may form a more compelling argumentation in future research as to how standardising mindfulness in face-composite instructions may be more acceptable by the general public, but for now, we assume that there is merit to embedding mindfulness instructions within eyewitness techniques and research trials.

CONFLICT OF INTEREST
The authors declare no conflicts of interest.

ENDNOTES
1 Although, note that some people find it uncomfortable to close their eyes, particularly in the presence of others (Nash et al., 2016).
2 Scored participant responses were subjected to Generalised Estimating Equations (GEE). This analysis contained one predictor, type of EvoFIT (coded as 1 = Mindfulness, and 2 = Standard, which was the reference category). The response variable (DV) was accuracy of composite naming (coded as described above: 0 = incorrect and 1 = correct response). The 'link' function specified was binary logistic (logit) to model the nominal (binary) responses. An 'exchangeable' structure was specified for the Working Correlation Matrix. Once built, the resulting parameter values [B and SE(B)] were checked to be within sensible limits (not too high or low), which might otherwise indicate an issue with fitting of the final model.

DATA AVAILABILITY STATEMENT
Data available from Kyriaki Giannou.