key: cord-0055411-765jgep1 authors: Karamaouna, Penny; Zouraraki, Chrysoula; Giakoumaki, Stella G. title: Cognitive Functioning and Schizotypy: A Four-Years Study date: 2021-01-08 journal: Front Psychiatry DOI: 10.3389/fpsyt.2020.613015 sha: cf764978483f981ecfba584f14d01daeb0fbfd61 doc_id: 55411 cord_uid: 765jgep1 Although there is ample evidence from cross-sectional studies indicating cognitive deficits in high schizotypal individuals that resemble the cognitive profile of schizophrenia-spectrum patients, there is still lack of evidence by longitudinal/follow-up studies. The present study included assessments of schizotypal traits and a wide range of cognitive functions at two time points (baseline and 4-years assessments) in order to examine (a) their stability over time, (b) the predictive value of baseline schizotypy on cognition at follow-up and (c) differences in cognition between the two time points in high negative schizotypal and control individuals. Only high negative schizotypal individuals were compared with controls due to the limited number of participants falling in the other schizotypal groups at follow-up. Seventy participants (mean age: 36.17; 70% females) were assessed at baseline and follow-up. Schizotypal traits were evaluated with the Schizotypal Personality Questionnaire. We found that schizotypal traits decreased over time, except in a sub-group of participants (“schizotypy congruent”) that includes individuals who consistently meet normative criteria of inclusion in either a schizotypal or control group. In these individuals, negative schizotypy and aspects of cognitive-perceptual and disorganized schizotypy remained stable. The stability of cognitive functioning also varied over time: response inhibition, aspects of cued attention switching, set-shifting and phonemic/semantic verbal fluency improved at follow-up. High negative schizotypy at baseline predicted poorer response inhibition and semantic switching at follow-up while high disorganized schizotypy predicted poorer semantic processing and complex processing speed/set-shifting. The between-group analyses revealed that response inhibition, set-shifting and complex processing speed/set-shifting were poorer in negative schizotypals compared with controls at both time points, while maintaining set and semantic switching were poorer only at follow-up. Taken together, the findings show differential stability of the schizotypal traits over time and indicate that different aspects of schizotypy predict a different pattern of neuropsychological task performance during a 4-years time window. These results are of significant use in the formulation of targeted early-intervention strategies for high-risk populations. Paul Meehl and Gordon Claridge have formulated the two dominant theories in the conceptualization of schizotypy, the quasi-dimensional and the fully-dimensional models, respectively. Meehl (1) proposed that schizotypy refers to a genetically mediated personality profile indicating liability to schizophrenia; when risk-factors for the development of the disease co-occur in individuals with elevated schizotypy, a percentage converts into the disease state. Claridge (2) pays particular attention to individual differences in schizotypal characteristics and considers those as traits lying in a continuum; only when exceeding a critical threshold, schizotypal traits indicate liability to schizophrenia, otherwise remaining part of what he describes as ". . . normal individual variation" [(2), p. 193]. Other reports also quite early suggested that schizotypy indicates proneness to schizophrenia (3) and schizophreniaspectrum personality disorders (4) . Since these early appraisals, (a) aspects of schizotypy have consistently been linked with higher rates of schizophrenia-spectrum traits and/or symptoms (5-7), (b) schizotypal personality has been further established as a key-element of the at risk mental state both in healthy high schizotypal individuals (8) and in individuals at clinical high risk for psychosis (9) , and (c) the connection between schizotypy and schizophrenia-spectrum disorders was established over the years at several endophenotypic levels (10) (11) (12) (13) (14) (15) (16) (17) . Most importantly, high schizotypy has been associated with transition into schizophrenia (18) (19) (20) (21) (22) . Schizophrenia is a complex disorder characterized by neuroanatomical (23, 24) , genetic (25, 26) , psychophysiological (27, 28) and functional (29, 30) impairments. Among these deficits, cognitive decline is a core feature of the disorder as evidenced by neuropsychological (31, 32) as well as functional neuroimaging (33, 34) studies. It is of note that cognitive decline is apparent as early as in the prodromal state (35, 36) and in the first episode (33, 37) of the disorder as well as in individuals at clinical (38, 39) or familial (38, 40) high-risk. To further strengthen the link between schizophrenia and schizotypy, poor cognitive functioning has also been well-established in high schizotypal individuals [for reviews and a meta-analysis see (10, 14, 41) ], although the severity of cognitive decline in this population is lower compared with schizophrenia patients (41) . Nevertheless, non-significant findings on the relationship between schizotypal traits and cognition have also been reported for several neurocognitive domains, such as set-shifting (42) (43) (44) (45) , problem solving (46), verbal fluency (46, 47), working memory (48), verbal memory (43, (47) (48) (49) (50) , processing speed (46, 51) , and aspects of attention (50, 52) . In a high percentage of studies, cognitive impairments are reported to remain stable over time in schizophrenic examinees (53, 54) . The temporal stability of schizotypy has been demonstrated with confirmatory factor analysis (55) . In the same line, schizotypal traits assessed with various versions of the Schizotypal Personality Questionnaire [SPQ; (56) ] have been reported (a) to have moderate stability estimates from early adolescence to early adulthood (57, 58) , with genetic factors explaining a significant percentage of the reported stability (57), (b) to either decrease, remain stable or increase over a 2 years period depending on their baseline levels in young adults (59) , (c) to follow a different trajectory over a 2-years period as negative schizotypy was more stable compared with positive schizotypy in young army conscripts (60) and (d) to remain stable over a 10-years period in psychotic patients and their unaffected siblings, except for disorganized schizotypy which decreased at the 10-years follow up assessment (61) . Studies with the Chapman Psychosis Proneness Scales (62) (63) (64) in healthy young adults revealed a decrease in physical anhedonia, perceptual aberration and magical thinking at an 18-months (65) and a 2years follow up assessment (66) compared with baseline along with moderate interclass correlation coefficients of all scores (66) . Despite the evidence for stability of schizotypal traits over time, evidence on the association of schizotypy with cognitive functioning in longitudinal or follow-up studies is still limited. To our knowledge, only two studies have thus far dealt with the issue. Thus, (a) Wang et al. (67) reported moderate stability of prospective and working memory deficits in a college sample scoring in the top 10% of the total score in the SPQ over a period of 6 months and (b) Cohen et al. (68) found that the cognitive performance of community participants with increased social anhedonia improved over a 3-years period, except for verbal or visual working memory and attentional vigilance, for which there was no change between the baseline and the followup assessments. In a previous study (69) we examined a wide range of cognitive functions in a large community sample stratified for schizotypal traits, as assessed with the detailed four-factor model of schizotypy (70) . The present study aimed to examine (a) the stability of cognitive task performance and schizotypal traits over a 4-years period, (b) the predictive value of baseline schizotypy on cognition after 4 years of the initial assessment of participants and (c) potential differences in cognitive functioning between two time points (i.e., baseline and follow-up assessment after 4 years) in high schizotypal individuals, as defined with the fourfactor model (i.e., high paranoid or negative or disorganized or cognitive-perceptual or control individuals). Due to lack of participants fulfilling the criteria in order to be included in any schizotypal group, the latter aim was limited only to the comparison between high negative schizotypals and controls. A total of 263 participants who had taken part in the Prefrontally Mediated Endophenotypes in the Schizophrenia Spectrum (PreMES) study were contacted for the 4-years followup assessment. Of those, 193 either declined to participate for several reasons or could not be traced (Supplementary Table 1) . Therefore, 70 healthy community participants (mean age ± SD: 36.17 ± 9.38; 21 males/49 females) were assessed both at baseline and follow-up. They were older, had more years of education and comprised fewer females (all p-values <0.05) compared with the drop-out group; there were no differences between participants in the follow-up assessment and those who dropped-out in any SPQ measure except for ideas of reference that were slightly higher in the individuals who dropped-out (Supplementary Table 2 ). Exclusion criteria were as per Karagiannopoulou et al. (69) while a medical history was taken again at follow-up to confirm that there was no health-related change in the participants. The medical history questionnaire was identical at baseline and follow-up assessments and comprised questions related to general health issues [e.g., "Have you visited a medical doctor for any reason during the last five (at baseline)/four (at follow-up) years?" "Are you taking any prescribed or over the counter medication now?" "Are you taking any substances that could be considered addictive, except nicotine and caffeine?"] and questions related specifically to mental health [e.g., "Have you visited a psychiatrist during the last five (at baseline)/four (at follow-up) years?"; Have you visited a psychologist during the last five (at baseline)/four (at follow-up) years, even for counseling required to resolve everyday problems?"; "Do you have any relatives suffering from a psychiatric/psychological disorder?"]; none of the included participants responded positively in any question both at baseline and follow-up. At baseline assessment, participants had been divided into schizotypal groups according to standard criteria (i.e., schizotypy scores falling in the upper 10% for only one schizotypal factor) derived by a normative sample in Greece (70) . In detail, (a) negative schizotypals were those with scores ≥18 in the negative schizotypy factor but with scores <14 in the paranoid, <7 in the cognitive-perceptual and <8 in the disorganized schizotypy factors; (b) paranoid schizotypals were those with scores ≥14 in paranoid schizotypy factor but with scores <18 in the negative, <7 in the cognitiveperceptual and <8 in the disorganized schizotypy factors; (c) cognitive-perceptual schizotypals were those with scores ≥7 in the cognitive-perceptual schizotypy factor but with scores <18 in the negative, <14 in the paranoid and <8 in the disorganized schizotypy factors; (d) disorganized schizotypals were those with scores ≥8 in the disorganized schizotypy factor but with scores <18 in the negative, <14 in the paranoid, and <7 in the cognitive-perceptual schizotypy factors; (e) controls were those who did not meet the criteria for any schizotypy factor. At follow-up, participants were also administered the SPQ (56) and the above-mentioned criteria were applied for their categorization into schizotypal groups a second time. Of the 70 tested volunteers, 54 fell in the same group (31 controls, two cognitive-perceptual, 20 negative, and one disorganized schizotypals-"schizotypy congruent") and another 16 "converted" into another group (Supplementary Table 3 Schizotypal traits were evaluated with the Greek version (70) of the SPQ (56) . The SPQ is a 74-dichotomous-item questionnaire and items are grouped into nine subscales (ideas of reference, social anxiety, odd beliefs/magical thinking, unusual perceptual experiences, eccentric/odd behavior, lack of close friends, odd speech, constricted affect, and suspiciousness) in analogy to the diagnostic criteria for Schizotypal Personality Disorder (SPD). The subscales are organized into four schizotypal factors (70) : negative (including suspiciousness, social anxiety, lack of close friends and constricted affect), paranoid (including ideas of reference, suspiciousness and social anxiety), cognitiveperceptual (including odd beliefs/magical thinking and unusual perceptual experiences), and disorganized (including eccentric/odd behavior and odd speech). Wisconsin Card Sorting Test [WCST; (71) ] Set-shifting was assessed with a computerized version of the WCST. The task consisted of four stimulus cards that varied along three dimensions (shape, color, and number) and a target card. Participants were asked to match the target card with one if the stimulus cards and feedback was provided after each selection. In the first part of the task, the first match was always scored as correct and the rule applied by the participant was the first sorting principle. After six consecutive correct responses, the sorting principle changed and participants were informed of this. The next match according to either of the two remaining sorting principles was scored as correct and as previously, after six consecutive correct responses, the sorting principle changed again and participants were informed. The third match was scored as correct only when the last sorting principle was applied. In the second part of the task, participants were required to repeat the three sorting principles in the same order. The task was discontinued when six categories were completed or when the target cards were exhausted. Outcome variables were (a) total number of completed categories, (b) number of unrelated matched cards, (c) perseverative errors either of Milner-or Nelson-type [Milner-type perseverative errors: responses that were correct on the immediately preceding stage of the test (72); Nelson-type perseverative errors: all other perseverative errors (71) ], and (d) Milner-and Nelson-non perseverative errors. Executive working memory was examined with the LNS. The Greek version (74) of the task was used. Strings of increasing difficulty comprising intermingled letters and numbers were read to the participants who were required to recite these strings after reordering of the information (i.e., in numeric and alphabetical order). The outcome variable was the total number of correct responses. Processing speed/set-shifting was evaluated with the Greek version (76) of TMT. The task consisted of two parts: in Part A, participants were required to connect 25 consecutively numbered circles, as quickly as possible. In Part B, participants were required to connect 25 consecutively numbered or lettered circles by alternating between the two sequences. The outcome variables were the seconds required by the participants to complete each part of the test. Verbal Fluency Test (77) The Greek version of the task was used for the assessment of phonemic and semantic fluency. In the phonemic fluency condition, participants were required to produce in 3 consecutive minutes as many as possible words beginning with the letter X (chi), S (sigma), and A (alpha) according to standard instructions. In the semantic fluency condition, participants were required to produce in 3 consecutive minutes, according to standard instructions, as many as possible words belonging to each of the following semantic categories: animals, fruit and objects. Outcome variables were (a) correct responses, (b) perseverative errors (i.e., words generated more than once), (c) intrusion errors (i.e., words that did not belong to the category required), (d) clusters (i.e., three or more consecutive words beginning with the same two letters and having the same sound or two consecutive words that differed only in a vowel sound or words that were homophones for the phonemic part; three or more consecutive words belonging to the same semantic sub-category in the semantic part), and (e) switches (i.e., total correct responses minus the number of words related to each cluster plus number of clusters) per sub-task. The task assesses abstract reasoning. It comprised five sets of 12 abstract patterns, each with one missing piece. Participants were required to select the choice that best matched the pattern out of the possible answer choices accompanying every pattern. Items within a set were of increasing difficulty. The outcome variable was the total number of correct selections. Attention Switch task (AST) assesses cued attentional switching. In every trial, an arrow appears on the left or the right half of the computer screen. A cue presented on the screen indicates whether the participant should make a response about the direction of the arrow or the side of the screen that the arrow was presented. A number of trials includes congruent stimuli (i.e., the arrow is on the right half of the screen and points to the right) whereas another number of trials include incongruent stimuli (i.e., the arrow is on the right half of the screen and points to the left), which require higher cognitive demands. Outcome variables were (a) congruency cost in mean correct responses (i.e., the difference between response latency of congruent vs. incongruent trials), (b) switch cost in mean correct responses (i.e., the difference between response latency of non-switched vs. switched trials), (c) total correct responses in switched and nonswitched trials, respectively, (d) total commission errors (i.e., total number of trials in which the examinee responded either before the end of the window or before the appearance of the stimulus) in switched and non-switched trials, respectively. Stop-Signal task (SST) examines response inhibition. In every trial, a white ring is presented in the center of the screen. After a delay of 500 ms, a white arrow pointing either to the right or to the left is presented within the ring. In the first part of the task, participants were required to press the right-hand button of a touch pad when the arrow points to the right or the left-hand button of the touch pad, when the arrow points to the left. In the second part of the task (five blocks with 64 trials each), participants were required to do as previously unless they hear an auditory stimulus. When they heard the auditory stimulus, participants were required to give no response. Outcome variables were (a) correct responses in the "stop condition" (i.e., when participants are required to give no response), (b) correct responses in the "go condition" (i.e., when participants are required to press either the right-or the left-hand button of the touch pad according to the direction of the arrow), (c) errors in the "stop condition" (i.e., when participants press the button of the touch pad that does not match the direction of the arrow in trials with the auditory stimulus), (d) errors in the "go condition" (i.e., when participants press the button of the touch pad that does not match the direction of the arrow in trials without the auditory stimulus), and (e) reaction time for correct response in the "go condition." Stockings of Cambridge [SoC; (80) ] examines planning and complex problem solving abilities. Participants were required to compare two different arrangements of "balls" in "socks" (one presented on the top half of the screen, the other at the bottom half) and re-arrange (with the minimum possible number of moves), the balls in the lower half in order to match the target arrangement in the upper half. The problems are of increasing difficulty. Participants were required to plan the complete sequence of moves needed prior to their first move. Outcome variables were (a) number of problems solved correctly with the minimum moves, (b) mean number of moves, (c) mean initial thinking time (i.e., the time taken to organize the solution of the problem prior to execution of the first move), and (d) mean subsequent thinking time (i.e., the time required for the achieving the solution to the problem). Spatial Working Memory [SWM; (80) ] was administered for the assessment of spatial working memory and strategy formation. Participants were required to search through an increasing number of boxes randomly arranged on the screen, until they find a token that, at any one time, is hidden in one of the boxes; upon finding the token, participants were required to place it in a "home area." The key instruction was that once a token has been found within a particular box, that box should never be used again to hide a token. On each trial, every box is used once to hide one token, such that the total number of tokens to be found equals to the number of boxes on the screen. Outcome variables were (a) between errors (i.e., times of re-visiting a box in which a token was previously found), (b) within errors (i.e., times of revisiting a box already found to be empty during the same search), (c) double errors (i.e., errors that can be categorized as both within and between errors), (d) strategy score (i.e., an efficient strategy is to follow a predetermined search sequence, beginning with a specified box and then return to start each new sequence with that same box as soon as a token has been found; a high score indicates poor strategy), and (e) mean search preparation time (i.e., the mean time between searches; for the first search it refers to the time between the on-screen presentation of the problem and the first touch of the examinee and for subsequent searches it refers to the time between the placement of token in the home area and the time of the next touch). Upon arrival at the laboratory, subjects self-rated their mood and feelings on a battery of 16-item visual analog scales [VAS; (81) ] of 10 cm each. The raw values for each item were weighted with their respective factor loading and the weighted values were allocated to "alertness, " "anxiety, " and "discontentment" factors (82) . The outcome variable was the average of the weighted values for each factor. The stability of schizotypal traits and neuropsychological task performance between baseline and follow-up assessments was examined with a series of repeated measures analyses of variance (ANOVAs) with time point (two levels) as the withinsubjects factor separately for the total sample (n = 70) and the "schizotypy congruent" group (n = 54). The rank order stability of participants' schizotypy scores was further examined with Pearson correlations between the baseline and followup scores separately for the total sample and the "schizotypy congruent" group. Associations between schizotypal traits at baseline and neuropsychological task performance at follow-up were examined with a series of stepwise regressions (dependent variable: neuropsychological measure; predictors: schizotypal factor scores; confounders: sex, age and smoking habits at baseline) separately for the whole sample and the "schizotypy congruent" group. To reduce the probability of type I error, we applied a Bonferroni correction (0.05/9 neuropsychological tasks = 0.0056); therefore, only p-values <0.0056 were considered as significant and p-values <0.05 were considered as trends for significance. In the "schizotypy congruent" sub-sample, betweengroup differences between controls (n = 31) and negative schizotypals (n = 20) in demographic variables (age, years of education, smoking habits), VAS and SPQ scores at followup were examined with parametric or non-parametric tests according to normality of the distribution; sex differences were examined with χ 2 analysis and differences in neuropsychological tasks performance were examined with 2 × 2 repeated measures ANOVAs with time point (baseline and follow-up) as the withinand group (control or negative schizotypals) as the betweensubjects factors. Due to the small sample size of the two groups, we considered the between-group differences as preliminary findings and we did not apply the Bonferroni correction in these analyses. In the analyses of the total sample, there was a significant main effect of time point for the cognitive-perceptual [F (1, 69) In all these measures, there was a decrease in the scores at follow-up ( Table 1 , upper panel). The Pearson's correlations between the factor scores of the two time points revealed that the strongest coefficient was that of negative schizotypy scores (r = 0.907, p < 0.001) followed by paranoid (r = 0.658, p < 0.001), cognitive-perceptual (r = 0.637, p < 0.001), and disorganized (r = 0.510, p < 0.001) schizotypy. At a sub-scale level, the strongest correlation was found for lack of close friends (r = 0.819, p < 0.001) followed by strong coefficients for constricted affect (r = 0.783, p < 0.001), odd speech (r = 0.730, p < 0.001), excessive social anxiety (r = 0.695, p < 0.001), eccentric behavior (r = 0.625, p < 0.001), and suspiciousness (r = 0.611, p < 0.001) and moderate coefficients for odd beliefs (r = 0.589, p < 0.001), ideas of reference (r = 0.487, p < 0.001), and unusual perceptual experiences (r = 0.465, p < 0.001). The correlation coefficient for the total SPQ score between baseline and follow-up was (r = 0.771, p < 0.001). A detailed description of the correlation matrix is provided in Supplementary Table 4 (upper panel). In the analyses of the "schizotypy congruent" group, though, a significant main effect of time point was found only for the cognitive-perceptual [F (1, 53) The Pearson's correlations between the factor scores of the two time points revealed that the strongest coefficient was that of negative schizotypy scores (r = 0.931, p < 0.001) followed by cognitive-perceptual (r = 0.705, p < 0.001), paranoid (r = 0.700, p < 0.001), and disorganized (r = 0.482, p < 0.001) schizotypy. At a sub-scale level, the strongest correlation was found for lack of close friends (r = 0.837, p < 0.001) followed by strong coefficients for constricted affect (r = 0.796, p < 0.001), eccentric behavior (r = 0.758, p < 0.001), excessive social anxiety (r = 0.718, p < 0.001), odd speech (r = 0.693, p < 0.001), suspiciousness (r = 0.630, p < 0.001), and odd beliefs (r = 0.608, p < 0.001) and moderate coefficients for ideas of reference (r = 0.460, p < 0.001) and unusual perceptual experiences (r = 0.457, p < 0.001). The correlation coefficient for the total SPQ score between baseline and follow-up was r = 0.891 (p < 0.001). A detailed description of the correlation matrix is provided in Supplementary Table 4 (lower panel). In the total sample, there was a significant main effect of time point for (a) SST correct responses in the "stop" [F (1, 69) In the whole sample, only older age was significantly associated [F (1,69) = 9.02, p < 0.005, R 2 = 0.117] with more total between errors (beta = 0.342, t = 3.003, p < 0.005) and prolonged mean search preparation time [F (1,69) = 9.50, p < 0.005, R 2 = 0.123; beta = 0.350, t = 3.082, p < 0.005]. In the "schizotypy congruent" group, a similar pattern was observed: older age was significantly associated [F (1,53) = 9.93, p < 0.005, R 2 = 0.160] with more total between errors (beta = 0.400, t = 3.150, p < 0.005) and longer mean search preparation time [F (1,53) = 11.85, p < 0.001, R 2 = 0.186; beta = 0.431, t = 3.442, p < 0.001]. In the whole sample, only high negative schizotypy tended to be associated [F (1, 69) = 5.31, p < 0.05, R 2 = 0.072] with fewer completed categories (beta = −0.269, t = −2.303, p < 0.05). In the "schizotypy congruent" group, we did not find any significant models (all p-values >0.05). In the whole sample, only older age tended to be associated [F (1, 69) = 5.95, p < 0.05, R 2 = 0.284] with more perseverative errors (beta = 0.284, t = 2.439, p < 0.05). In the "schizotypy congruent" sub-sample, female sex was (a) associated [F (1, 53) = 8.69, p < 0.005, R 2 = 0.143] with correct responses (beta = 0.378, In the whole sample, (a) older age was associated [F (1, 69) There were no differences in any demographic variables or VAS scores between the control and the negative schizotypal groups (all p-values >0.07). The negative schizotypal group, though, had higher total SPQ, paranoid, negative and disorganized factor scores (all p-values <0.005). A detailed description is provided in Table 3 . The descriptives of the two groups' performance in the tasks with either between-group differences or interactions involving group are presented in Table 4 ; the descriptives of the remaining tasks are presented in Supplementary Table 5 . Significant main effects of group and time point were found for the correct responses in the "stop" (Figure 1, upper panel) . No other significant main effects or interactions were found (all p-values >0.055). The negative schizotypal group had poorer performance in the second part of the task at both time points [main effect of group: F (1,49) = 6.83, p < 0.05; η 2 = 0.122]. We did not find any other significant main effects or interactions (all p-values >0.095]. Significant main effects of time point were found for correct responses [F (1, 49) = 10.76, p < 0.005; η 2 = 0.180] and number of switches [F (1, 49) = 13.14, p < 0.001; η 2 = 0.211] as both groups gave more correct responses and made more switches at followup compared with the baseline assessment. No other significant main effects or interactions were revealed (all p-values >0.062). A significant group × time point interaction was revealed for the number of switches [F (1,49) = 6.20, p < 0.05; η 2 = 0.112] according to which, the negative schizotypal group made fewer switches at follow-up compared with the baseline assessment, while the control group presented with the opposite pattern (Figure 1, lower panel) . We also found a significant main effect of time point for the number of correct responses [F (1, 49) = 8.67, p < 0.005; η 2 = 0.150] with both groups improving at follow-up. The analyses did not reveal any other significant main effects or interactions (all p-values >0.055). As regards schizotypal traits, a different pattern emerged when examining the total sample and the sub-sample of "schizotypy congruent" individuals (i.e., participants who at follow-up fulfilled the criteria of inclusion in the same schizotypal group as in the baseline assessment) separately. Thus, in the total sample we found that schizotypal traits did not remain stable over a 4years period, as a decrease in the scores of all schizotypal factors, SPQ total score and the majority of sub-scale scores was found at follow-up. Contrary to this, in the "schizotypy congruent" subgroup, negative schizotypy (and all its constituents) remained unchanged, as did ideas of reference (which play a central role in paranoid schizotypy), unusual perceptual experiences (which is a core feature of cognitive-perceptual schizotypy) and odd speech (one of the two characteristics of disorganized schizotypy). The decrease in schizotypal traits at follow-up is in accordance with findings indicating fluctuations in SPD-like features (59, 83, 84) or schizotypal traits per se (61, 65, 66) over time. Nevertheless, certain schizotypal traits (i.e., the negative schizotypal dimension, ideas of reference, unusual perceptual experiences and odd speech) seem to have a more enduring nature in certain individuals (i.e., those falling in the extreme ends in the continuum of schizotypy). Plausible explanations for these findings are based on the facts that negative schizotypy is the sub-clinical analog of negative symptoms, which have been reported to be a stable feature of schizophrenia (85, 86) . Further supporting the association of negative schizotypy with negative symptoms, Cohen et al. (84) recently reported that individuals with increased social anhedonia presented with increased negative symptom characteristics over a 3-years period. With respect to the other schizotypal traits that remained stable, unusual perceptual experiences and ideas of reference have also been reported to show stability over time in SPD (87) as is the case with odd/disorganized speech in schizophrenia (88) . The correlational analyses aimed to further explore the rank order stability of participants' scores, i.e., ". . . how individuals maintain their standing on a trait level compared to others in a population over time" [(89), p. 27]. We found that the rank order stability of negative, paranoid and cognitive-perceptual factor scores ranged from very high to high while disorganized schizotypy was moderately rank-ordered; both groups also had a similar pattern of rank order stability at a sub-scale level. This is the first study to report findings on the rank order stability of schizotypal traits with the four-factor model and it is especially interesting that the reported findings are in accordance with the schizophrenia literature indicating high rank order stability of negative symptoms (90) as well as findings suggesting that aspects of paranoid and cognitive-perceptual schizotypy are the most prevalent and highly rank ordered features of SPD (87) . In accordance with previous studies that have employed the same neuropsychological tasks as in the present one and have examined the stability of cognitive functioning (91-96), we found that there was no difference in planning/problem solving, spatial working memory/strategy formation, processing speed, executive working memory and abstract reasoning between the two time points. On the other hand, response inhibition and aspects of cued attention switching, set-shifting as well as phonemic and semantic verbal fluency improved at follow-up compared with baseline, although the effect of time on set-shifting was abolished in the "schizotypy congruent" sub-group probably due to the smaller sample size. Even though there are findings supporting the stability of these cognitive functions over time (93, (97) (98) (99) , there is also evidence indicating that performance in the tasks employed here is subject to practice effects resulting in improved performance (99) (100) (101) (102) (103) . The majority of these studies include assessments at shorter time-intervals than in the present one; however, there are findings supporting the persistence of these effects at time intervals comparable to the 4-years of our follow-up assessment (104) (105) (106) . This is the first study to explore the predictive significance of schizotypal traits on cognitive functions after a 4-years period. Moreover, we examined the afore-mentioned associations in our total sample and in the sub-sample of "schizotypy congruent" individuals, separately. It was interesting to find out that a different pattern of associations was revealed between different schizotypal dimensions or specific schizotypal traits in the two groups. High negative schizotypy at baseline predicted poorer response inhibition (as examined with more errors in the most "inhibition-demanding" condition of the SST task) and poorer semantic switching (i.e., shifting to another semantic category in order to produce more correct responses) 4 years later both in the total sample and in the "schizotypy congruent" group. Ettinger et al. (107) have reported that negative schizotypy is associated with poor response inhibition cross-sectionally, in analogy to findings showing associations between negative schizophrenia symptoms with both response inhibition (108) and semantic switching (109) . The present findings, therefore, suggest that the negative effect of schizotypy on these two executive processes also remains stable over time. Response inhibition and semantic switching are mediated by a frontaltemporal-parietal network in healthy individuals (110) (111) (112) (113) and in schizophrenia patients (114) (115) (116) . Interestingly, alterations within this network seem to be of central importance in SPD as reduced gray matter volume (117, 118) has been reported in SPD patients along with associations between cortical thinning (119) or gray matter volume reductions (118) with SPD symptoms. Although the literature on schizotypy is still limited, there is evidence implicating this neural network in negative schizotypy (120, 121) , as well. Therefore, the critical link between all implicated constructs (i.e., negative schizotypy, response inhibition, semantic switching, schizophrenia, SPD) seems to be the neural circuitry connecting the frontal, temporal, and parietal lobes. Although the methodology of the present study allows only for indirect conclusions, we propose that inefficient functioning of/processing within this network is a persistent feature of negative schizotypy identifying sub-optimal cognitive functioning mediated by this network over time. High disorganized schizotypy at baseline predicted more intrusion errors and tended to predict fewer correct responses in the semantic fluency task only in the total sample. Tan and Rossell (122) have already reported reduced semantic fluency productivity as disorganized schizotypy increases. In the present study, though, we found a stronger predictive value of disorganized schizotypy for semantic intrusion errors (i.e., production of words belonging to a semantically different category). Correct responses potentially reflect semantic information reserve capacities, while semantic intrusion errors could be considered as failures in semantic processing either due to poor strategic search of representations classified according to their meaning or misattribution of meaning to representations. It is of note, that disorganized schizotypy has been associated with a wide neural network encompassing several brain structures, such as the superior longitudinal fasciculus (121), hippocampus (123) , and superior temporal gyrus (124) , that are crucial for semantic processing (125) (126) (127) (128) (129) (130) (131) , thus proving a plausible explanation for our finding. The fact that the association of disorganized schizotypy and semantic processing was abolished in the "schizotypy congruent" sub-group is most probably to the smaller sample size of this group resulting in limited variation of semantic fluency intrusion errors that is required for associational analyses. High disorganized schizotypy at baseline also predicted poorer complex processing speed/set-shifting, as assessed with the multitasking second part of TMT, both in the total sample and "schizotypy congruent" individuals. In support of this finding, cross-sectional studies have revealed that disorganized schizotypy correlates with response inhibition in relatives of schizophrenia patients as well as controls (107) and disorganization symptoms in schizophrenia patients have been associated with TMT performance (132, 133) and other measures of response inhibition (134) . At a neuroanatomical level, the neural substrate underlying performance in TMT includes a wide frontal-temporal-parietal network with a frontal cluster including the cingulate and insular cortices and the frontal gyrus acting as a central node (135) . Disorganized schizotypy has been associated with reduced volume of the insula (124), cingulate cortex and frontal gyrus (121, 124) as well as altered insular information processing (136) , and thinning of the anterior cingulate (137) . Taken together, our findings suggest that the association of disorganized schizotypy with cognitive functioning relies on different mechanisms and holds out over time: a frontaltemporal network mediates the relationship of disorganized schizotypy with semantic processing while a cluster of frontal regions is the key-neural substratum for connecting disorganized schizotypy with complex processing speed/set-shifting. The follow-up assessment did not include participants falling in all schizotypal dimensions due to the high-rate of drop-out; thus, between-group differences were limited between negative schizotypals and controls and due the between-group differences in most schizotypal dimensions, we could assume that the former group has an overall "heavier dose" of schizotypal traits compared with the latter. As regards the effects of time and in accordance with the analyses of the total sample, neuropsychological task performance in both groups improved at follow-up in measures of response inhibition, set-shifting, phonemic and semantic fluency and cued attention switching. The analyses also revealed that the group of negative schizotypals had poorer response inhibition (as assessed with the SST), set-shifting (as assessed with the completed categories in the WCST), and complex processing speed/set-shifting (examined with the second part of TMT) compared with controls at both time points. These findings (a) cannot be attributed to differences in the demographic characteristics of the participants as there were no betweengroup differences in these variables, (b) are in accordance with previous cross-sectional studies (69, 107, 138) , (c) highlight the persistent nature of specific inefficiencies in cognition in negative schizotypy, and (d) supplement the association of this schizotypal dimension with processes underlaid by the frontal-temporalparietal network described in the previous section. Interesting group × time point interactions were also revealed in two fundamental measures: the negative schizotypal group was more prone to failure to maintain set (139) as indicated by the selection of more unrelated cards in the WCST (i.e., choosing a card that has no common features with the target stimuli after acquiring the sorting principle) and made fewer semantic switches at follow-up compared with baseline performance; the opposite pattern was observed in controls. Maintaining set and switching are inter-related as they both require efficient self-monitoring, response inhibition and vigilance. They are also primary features of successful performance in the WCST and semantic fluency tasks [e.g., reported correlations between failure to maintain set and completed categories as well as perseverative errors (140) and between semantic switching and word production (141, 142) ]. With respect to the present findings, it is tempting to propose that maintaining set and semantic switching in negative schizotypals are progressively deteriorating thus leading to increased effort for the successful completion of the tasks. These findings, however, are preliminary and the aforementioned suggestion should be viewed with greater caution as far as maintaining set is concerned due to the reported low test-retest reliability of this measure (143, 144) . Taken together, the findings of the present study indicate that schizotypal traits, when analyzed with a detailed four-factor model, decrease over a 4-years period in the general population. The exception is negative schizotypy and odd speech, which is a central feature of disorganized schizotypy; these traits remained stable, in accordance to the literature on schizophrenia and spectrum disorders. The stability of cognitive functioning also varied over time. Thus, response inhibition, aspects of cued attention switching, set-shifting as well as phonemic and semantic verbal fluency are more likely to improve, possibly due to persistent learning/practice effects when performing the same test twice. The need for alternate forms of neuropsychological tasks has already been highlighted (145, 146) and the present study suggests that this applies not only in clinical but also in research settings. High negative schizotypy at baseline predicted poorer response inhibition and semantic switching at follow-up, further supporting and supplementing the involvement of a frontaltemporal-parietal network in all these latent constructs. High disorganized schizotypy at baseline predicted poorer semantic processing and complex processing speed/set-shifting. A frontaltemporal network is suggested to mediate the association of disorganized schizotypy with semantic processing while a cluster of frontal regions is suggested as the key-element for its connection with complex processing speed/set-shifting. The association of negative schizotypy with cognition over time was further explored with preliminary findings on differences between negative schizotypal and control individuals. Poor response inhibition, set-shifting, and complex processing speed/set-shifting were found to be consistently impaired in the former group, as they performed lower than controls in both time points. The ability to maintain set and semantic switching, though, were found to be progressively deteriorating in negative schizotypy. The present findings highlight the importance of taking into consideration the differences in schizotypal traits when designing early-intervention programs for high-risk populations. The fact that different association patterns with cognition were revealed, further advances the formulation of more targeted approaches depending on the prevailing schizotypal traits not only in the schizophrenia spectrum but also in other clusters of mental disorders; for example features of negative schizotypy have been associated with aspects (147) (148) (149) or the prevalence rates (5) of mood disorders. To our knowledge, this is the first study exploring associations of schizotypal traits with cognitive functioning over a 4-years period. It is of note that the participants were community residents covering a wide age-range instead of the most commonly included college students' samples that have been reported to under-represent the general population (150) . For the assessment of schizotypy we applied the detailed fourfactor model, which allows for more thorough delineations of schizotypal traits. We also applied strict normative criteria for the identification of negative schizotypal and control individuals instead of dividing the participants according to a samplewise approach (e.g., by median or percentile splits in the current sample). Nevertheless, there are certain limitations that should be taken into account. First, we had a quite high drop-out rate in the initial sample resulting in a small sample size at follow-up that also differed in demographics compared with the drop-out participants. This also resulted in a small number of "schizotypy incongruent" individuals that were not examined at all at followup. Second, we assessed schizotypy only with a self-report scale, which is widely used but was recently reported not to be fully concordant with interview-based assessments (151). Third, although we examined the subjective state of mood and feelings on the day of testing, we did not include these data in our regression analyses (this would increase the complexity of our models to a level that could not be justified by the current sample sizes) and we did not examine other factors (e.g., anxiety or discontentment on a daily basis) that might have interfered with the participants' performance, especially since the study was conducted during the COVID-19 pandemic. The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation. The studies involving human participants were reviewed and approved by the Research Ethics Committee of the University of Crete. The patients/participants provided their written informed consent to participate in this study. PK collected the data, conducted the initial analyses, and wrote the first draft of the manuscript. CZ collected the data and supplemented the manuscript. SG designed the study and supplemented the statistical analyses and the manuscript. Fully and quasi-dimensional constructions of schizotypy Tracking the taxon: on the latent structure and base rate of schizotypy The latent structure of schizotypy: I. Premorbid indicators of a taxon of individuals at risk for schizophrenia-spectrum disorders Social anhedonia and schizotypy in a community sample: the Maryland Longitudinal Study of Schizotypy Prediction of psychopathology and functional impairment by positive and negative schizotypy in the Chapmans' Ten-Year Longitudinal Study Barrantes-Vidal N. Prediction of prodromal symptoms and schizophrenia-spectrum personality disorder traits by positive and negative schizotypy: a 3-year prospective study Distress and metacognition in psychosis prone individuals: comparing high schizotypy to the at-risk mental state Comorbid personality disorders in individuals with an at-risk mental state for psychosis: a meta-analytic review Cognitive and prepulse inhibition deficits in psychometrically high schizotypal subjects in the general population: relevance to schizophrenia research Emotion processing deficits in the different dimensions of psychometric schizotypy Evidence of a dimensional relationship between schizotypy and schizophrenia: a systematic review Genetics, cognition, and neurobiology of schizotypal personality: a review of the overlap with schizophrenia. Front Psychiatry Cognition and brain function in schizotypy: a selective review Schizotypal personality disorder: a current review The role of schizotypy in the study of the etiology of schizophrenia spectrum disorders The concept of schizotypy-a computational anatomy perspective Developing psychosis and its risk states through the lens of schizotypy Psychosis-predictive value of self-reported schizotypy in a clinical high-risk sample Improved individualized prediction of schizophrenia in subjects at familial high risk, based on neuroanatomical data, schizotypal and neurocognitive features Psychotic conversion of individuals at ultra-high risk for psychosis: the potential roles of schizotypy and basic symptoms A developmentally-stable pattern of premorbid schizoid-schizotypal features predicts psychotic transition from the clinical high-risk for psychosis state Magnetic resonance imaging and the prediction of outcome in firstepisode schizophrenia: a review of current evidence and directions for future research Neuroimaging in schizophrenia Genetics of schizophrenia: a consensus paper of the WFSBP task force on genetics Unravelling the genetic basis of schizophrenia and bipolar disorder with GWAS: a systematic review A review of endophenotypes in schizophrenia and autism: the next phase for understanding genetic etiologies Sensorimotor gating deficits in "Two-Hit" models of schizophrenia risk factors Assessing the relationship between performance on the University of California Performance Skills Assessment (UPSA) and outcomes in schizophrenia: a systematic review and evidence synthesis A systematic review of long-term outcomes of patients with psychosis who received early intervention services Neurocognitive impairment in deficit and non-deficit schizophrenia: a meta-analysis A meta-analysis of executive dysfunction in patients with schizophrenia: different degree of impairment in the ecological subdomains of the Behavioural Assessment of the Dysexecutive Syndrome Neural correlates of cognitive deficits across developmental phases of schizophrenia Neuroimaging studies of cognitive function in schizophrenia The longitudinal course of cognitive impairment in schizophrenia: an examination of data from premorbid through posttreatment phases of illness Cognitive development prior to onset of psychosis Meta-analysis of cognitive impairment in first-episode bipolar disorder: comparison with first-episode schizophrenia and healthy controls Cognitive deficits in youth with familial and clinical high risk to psychosis: a systematic review and meta-analysis Neurocognitive dysfunction in subjects at clinical high risk for psychosis: a meta-analysis Neurocognition in youth and young adults under age 30 at familial risk for schizophrenia: a quantitative and qualitative review Neuropsychological correlates of schizotypy: a systematic review and meta-analysis of cross-sectional studies Individual differences in spatial working memory in relation to schizotypy Emotional intelligence and social functioning in persons with schizotypy Schizotypy and genetic loading for schizophrenia impact upon neuropsychological status in bipolar II and unipolar major depressive disorders Visual information processing deficits as biomarkers of vulnerability to schizophrenia: an eventrelated potential study in schizotypy Lenzenweger MF, Gold JM. Auditory working memory and verbal recall memory in schizotypy Theory of mind, neurocognition, and functional status in schizotypy Schizotypal traits and cognitive function in healthy adults Executive function deficits associated with symptoms of schizotypy and obsessive-compulsive disorder Inhibition and facilitation in schizotypy Cognitive deficits in psychotic disorders: a lifespan perspective Cognitive impairment in psychotic illness: prevalence, profile of impairment, developmental course, and treatment considerations The Chapman psychosis-proneness scales: consistency across culture and time The SPQ: a scale for the assessment of schizotypal personality disorder based on DSM-III-R criteria Heritability and longitudinal stability of schizotypal traits during adolescence The stability of schizotypy across time and instruments Developmental trajectories of schizotypal personality disorder-like behavioural manifestations: a two-year longitudinal prospective study of college students Psychometric properties of the greek version of the Schizotypal Personality Questionnaire (SPQ) in young male obligatory conscripts: a two year test-retest study Ten-year stability of self-reported schizotypal personality features in patients with psychosis and their healthy siblings Scales for physical and social anhedonia Body-image aberration in schizophrenia Magical ideation as an indicator of schizotypy Trajectories of schizotypy and their emotional and social functioning: an 18-month followup study Two-year stability of psychosis proneness scales and their relations to personality disorder traits Stability of prospective memory deficits in individuals with schizotypal personality traits Neuropsychological functioning and social anhedonia: three-year follow-up data from a longitudinal community high risk study Cognitive profiles of schizotypal dimensions in a community cohort: common properties of differential manifestations The validity of the Schizotypal Personality Questionnaire in a Greek sample: tests of measurement invariance and latent mean differences A modified card sorting test sensitive to frontal lobe defects Effects of different brain lesions on card sorting Wechsler Adult Intelligence Scale-Fourth Edition (WAIS-IV) Wechsler Adult Intelligence Scale, Standardization in Greek Trail Making Test A and B: normative data stratified by age and education A normative study of the trail making test A and B in Greek adults The verbal fluency task in the Greek population: normative data, and clustering and switching strategies Manual for Raven's Progressive Matrices and Vocabulary Scales. Section 1: General Overview A study of performance on tests from the CANTAB battery sensitive to frontal lobe dysfunction in a large sample of normal volunteers: implications for theories of executive functioning and cognitive aging Planning and spatial working memory following frontal lobe lesions in man The action of sedatives on brain stem oculomotor systems in man The use of analogue scales in rating subjective feelings Two-year stability and change of schizotypal, borderline, avoidant, and obsessive-compulsive personality disorders Social anhedonia and clinical outcomes in early adulthood: a three-year follow-up study within a community sample The course of negative symptoms in first-episode schizophrenia and its predictors: a prospective two-year follow-up study Understanding the course of persistent symptoms in schizophrenia: longitudinal findings from the pattern study Two-year prevalence and stability of individual DSM-IV criteria for schizotypal, borderline, avoidant, and obsessive-compulsive personality disorders: toward a hybrid model of axis II disorders A prospective longitudinal 10-year study of schizophrenia's three major factors and depression Developmental course of personality disorder traits in childhood and adolescence Stability in schizophrenia symptoms over time: findings from the Mount Sinai Pilgrim Psychiatric Center Longitudinal Study The Raven's progressive matrices: change and stability over culture and time Stability of set-shifting and planning abilities in patients with schizophrenia Memory and executive impairment in schizophrenia: comparison with frontal and temporal brain damage Characterization of neurophysiologic and neurocognitive biomarkers for use in genomic and clinical outcome studies of schizophrenia Relative stability of neurocognitive deficits in early onset schizophrenia spectrum patients The trajectory of cognitive functioning following first episode mania: a 12-month follow-up study Executive functioning in schizophrenia spectrum disorder patients and their unaffected siblings: a ten-year follow-up study Reliability and plasticity of response inhibition and interference control No evidence for the reduction of task competition and attentional adjustment during task-switching practice Practice effects on the Wisconsin card sorting test-64 card version across 12 months The reliability of cluster and switch scores for the controlled oral word association test Test-retest reliability of measures commonly used to measure striatal dysfunction across multiple testing sessions: a longitudinal study Examining 3-month test-retest reliability and reliable change using the Cambridge Neuropsychological Test Automated Battery Practice and drop-out effects during a 17-year longitudinal study of cognitive aging Age and ability affect practice gains in longitudinal studies of cognitive change Effects of age and ability on components of cognitive change Response inhibition and interference control: effects of schizophrenia, genetic risk, and schizotypy Increased intra-individual variability of cognitive processing in subjects at risk mental state and schizophrenia patients Category fluency in schizophrenia: opposing effects of negative and positive symptoms? Neural systems supporting lexical search guided by letter and semantic category cues: a self-paced overt response fMRI study of verbal fluency Neural processes of preparatory control for stop signal inhibition Recruiting the right hemisphere: Sex differences in inter-hemispheric communication during semantic verbal fluency Dynamic reconfiguration of functional brain networks supporting response inhibition in a stop-signal task Effect of retrieval effort and switching demand on fMRI activation during semantic word generation in schizophrenia Altered brain network dynamics in schizophrenia: a cognitive electroencephalography study Shared and differential cortical functional abnormalities associated with inhibitory control in patients with schizophrenia and bipolar disorder Smaller neocortical gray matter and larger sulcal cerebrospinal fluid volumes in neuroleptic-naive women with schizotypal personality disorder Globally and locally reduced MRI gray matter volumes in neuroleptic-naive men with schizotypal personality disorder: association with negative symptoms Severity of cortical thinning correlates with schizophrenia spectrum symptoms Higher prefrontal cortical thickness in high schizotypal personality trait A multimodal imaging study of brain structural correlates of schizotypy dimensions using the MSS Disorganised schizotypy is selectively associated with poorer semantic processing in non-clinical individuals Anterior vs posterior hippocampal subfields in an extended psychosis phenotype of multidimensional schizotypy in a nonclinical sample Identifying grey matter changes in schizotypy using partial least squares correlation Semantic fluency deficits and reduced grey matter before transition to psychosis: a voxelwise correlational analysis Neuroanatomical correlates of verbal fluency in early Alzheimer's disease and normal aging Structural connectivity subserving verbal fluency revealed by lesion-behavior mapping in stroke patients Anatomy and white matter connections of the inferior frontal gyrus Cognitive speed and white matter integrity in secondary progressive multiple sclerosis Dissociating frontal and temporal correlates of phonological and semantic fluency in a large sample of left hemisphere stroke patients Neural correlates of verb fluency performance in cognitively healthy older adults and individuals with dementia: a pilot fMRI study Cognitive patterns in subtypes of schizophrenia Correlates of cognitive impairment in first episode schizophrenia: the EUFEST study Methodological considerations regarding the association of Stroop and verbal fluency performance with the symptoms of schizophrenia Neural signatures of trail making test performance: evidence from lesion-mapping and neuroimaging studies Spatio-temporal source cluster analysis reveals fronto-temporal auditory change processing differences within a shared autistic and schizotypal trait phenotype Developmental trajectories of cortical thickness in relation to schizotypy during adolescence WCST performance and schizotypal features in the first-degree relatives of patients with schizophrenia Failure to maintain set: a measure of distractibility or cognitive flexibility? Wisconsin card sorting test scores and clinical and sociodemographic correlates in schizophrenia: multiple logistic regression analysis Component analysis of verbal fluency performance in younger participants with moderate-to-severe traumatic brain injury Sex differences in verbal fluency: the role of strategies and instructions Test-retest reliability of the Wisconsin card sorting test in people with schizophrenia The reliability of the Wisconsin card sorting test in clinical practice Evidence-based indicators of neuropsychological change in the individual patient: relevant concepts and methods The robust reliability of neuropsychological measures: meta-analyses of test-retest correlations On the specificity of positive emotional dysfunction in psychopathology: evidence from the mood and anxiety disorders and schizophrenia/schizotypy Differential associations between schizotypy facets and emotion traits Association of positive, negative, and disorganized schizotypy dimensions with affective symptoms and experiences Neurocognition in psychometrically defined college schizotypy samples: we are not measuring the "right stuff Comparison of self-report and clinician-rated schizotypal traits in schizotypal personality disorder and community controls All authors contributed to the article and approved the submitted version. The authors wish to thank the participants for their help with the study. The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fpsyt. 2020.613015/full#supplementary-material Conflict of Interest: The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.Copyright © 2021 Karamaouna, Zouraraki and Giakoumaki. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.