Central Coherence: Is it a single construct, with a relationship to mentalising? The past 20 years has seen much interest in the development of cognitive profiles and mentalising ability, particularly in how they may account for some characteristics of autism spectrum disorders (ASDs).
Two of the most influential theories to date are: (i) Weak Central Coherence theory (WCC), which posits that those with WCC focus on detailed (local) features and fail to apply a global context in understanding their environment (Frith & Happe, 1994); and (ii) Theory of Mind (ToM), also referred to as ‘mentalising’, which relates to the ability to recognise ones’ own or others’ mental states (Baron-Cohen, Wheelwright, Hill, Rast & Plumb, 2001). Evidence of ToM impairments in individuals with ASD is well documented in the literature (Baron-Cohen et al. ,2001; Beaumont & Newcombe, 2006; Loth, Gomez & Happe, 2008; Happe & Frith, 2006).
Whilst some earlier findings were criticised for using measures claimed to have psychometric problems and/or ceiling effects, recent years has seen more advanced measures being introduced, such as the revised Reading the Mind in the Eyes Test (RET) (Baron-Cohen et al. , 2001). Improved measures have supported previous results. Baron-Cohen’s (2001) revised RET tests both an adults’ and childs’ individual ability to infer mental states from pictures of people’s eyes. In many experiements it has shown the ability to attribute mental states is impaired in both children and adults with ASD (Beaumont & Newcombe, 2006; Loth et al. 2008), in addition to identifying gender differences showing neurotypical males to be lower scorers than females (Baron-Cohen et al. , 2001). The theory that Central Coherence (CC) deficits are linked to ASD also has much empirical support, helping explain why individuals with ASD have a greater skill for focusing on smaller (local) details of an event and less susceptibility for visual illusions (Burnette, Mundy, Meyer, Sutton, Vaughan, Charak, 2005; Deruelle, Rondan, Gepner & Faggot, 2006; Happe & Frith, 2006).
This recognised tendency to have a bias for either local or global cognitive processing styles has also been found in neurotypical adults, leading some to claim (Happe, Briskman & Frith, 2001) that CC competence must exist on a single continuum, with local and global biases polarised. Measures of CC have commonly included tasks requiring attention to internal (local) details, such as tasks which require the neglect of canonical arrangement of dots when counting, a Homograph Reading Task (Burnette et al. 2005) and the Embedded Figures Tasks (EFT) (Baron-Cohen & Hammer, 1997). More recently, a measure focusing on biases towards global processing rather than local was conducted using hierarchical (global / local) Navon figures (Deruelle et al. , 2006). Findings were still found to support those of local-focused measures. Given the empirical evidence showing the strong relationships both WCC and ToM independently have with explaining deficiencies in people with ASD, it is not surprising that subsequent research exploring the relationship between WCC and ToM has followed.
An analysis of past research examining this issue was conducted and identified contrasting results. Interestingly, findings ranged from support of a significant correlation between the two cognitive anomalies (Baron-Cohen & Hammer, 1997; Jarrold, Butler, Cottington & Jimenez, 2000; Loth et al. , 2008), to moderate correlations (Burnette et al. , 2005), to no evidence of a correlation (Beaumont & Newcombe, 2006; Biao, Zhen & Ciping, 2006).
Raising further questions, particularly around the relationship among adults, is the fact that some evidence in support of correlations can not reliably be extrapolated to adults either due to inadequate testing measures (Baron-Cohen and Hammer, 1997) or testing having been conducted on children only (Jarrold et al. , 2006). As the relationship between WCC and ToM in adults is still not fully understood, the aim of the present study is to provide more conclusive evidence in this area.
In doing so, it has sought to use a testing measure which accounts for the limitations of EFT through a tool which allows for both local and global biases to be measured independently. Only through this process is it believed that certainty around whether the two biases exist in a unidimensional or multidemsional model can be ascertained. Based on the ability of Deruelle et al. ’s (2006) Navon Figure Task to provide a context which allows for each processing type to be explicitly defined, a similar version of this test has been used in this experiment.
Unlike the previous test however, adaptations have been made to allow for both local and global processing biases to be tested independently, as well as overcome the motor distractions which previously inhibited the use of response times. In addition to CC measures, an evaluati on of ToM using a RET measure has been incorporated to enable an unambiguous interpretation of CC’s relationship with mentalising. Based on the contradictions of past findings around a WCC and ToM relationship, as well as inclusive claims around CC’s unidimension, this study addressed three hypotheses.
Firstly, if there is indeed a relationship between the two cognitive states, as asserted by some of the previous research (Baron-Cohen and Hammer, 1997; Jarrold et al. , 2000), it is predicted that this will be represented by those with larger local bias scores being poorer at mentalising, reflected by a lower RET score. That is, there will be a negative correlation between local bias and RET scores. Secondly, if a single continuum model is the case, then it should follow that this experiment will show a strong negative correlation between local bias and global bias.
Furthermore, evidence of a significant positive correlation between Global Bias and RET should also be seen by those with a global processing bias having developed better mentalising skills, thus performing well in their RET scores. To examine these issues, evaluations were made of the linear relationships between local bias and RET scores, local and global bias, and global bias and RET scores using Pearson’s correlation. Method Participants Participants were neurotypical adults with good vision (with or without glasses), chosen by students of a second year psychology class at University of New England.
All participated voluntarily after reading an information sheet and giving their signed consent. Data was collected from 174 participants, 44 of which were then randomly chosen for analysis in this report. This included 22 females and 22 males aged 18 to 66 years. with a mean age of 34 years. Materials Two pencil and paper tests were conducted by each respondent, including Baron-Cohen’s revised Reading Eye Task (Baron-Cohen et al. , 2001) and the Navon Figures Task (Deruelle et al. , 2006). Procedure
Each participant was tested individually in a private residence, while seated at a comfortable table in an environment of good light. Task order was consistent across participants, starting with the RET, then the Congruent Navon task, followed by the Incongruent(Local), then the Incongruent(Global). For the Reading Eye Task, participants were shown a series of 36 photographs of the eye region, each with four emotional state adjectives. Participants were told to look at each set of eyes in turn, and choose which state best described each facial expression.
As they responded, the researcher ticked next to each target item if answered correctly. The measured variable was how many items were correctly answered. For the Navon Figure Task, participants were given the three Navon sheets in turn, each comprising of local elements (letters) organised to form 96 randomly ordered larger global letters. Sheets were placed at a distance of approximately 500m from participants’ eyes and for each sheet participants were given 60 seconds to verbally name as many of the letter figures as they could, as rapidly as they could.
Figures were read horizontally from left to right, and repeated from the top left figure in instances where participants reached the bottom right figure before the time limit was over. As participants read out figures correctly, the researcher placed a tick next to the target item. The measured variable was the number of ticks at the end of 60 seconds. Results For CC measures, the CC bias variables were generated by dividing each participants’ Incongruent(Local) score by their Congruent score (Local Bias), and their Incongruent(Global) score by their Congruent score (Global Bias).
Median values for the local and global ratios were calculated and data was assigned into one of two groups based on its comparative value to the median score. Inferential statistical tests were evaluated against an alpha level of . 05. Results showed no evidence of a significant relation between age and number of correct RET responses (r(44)=. 30,p>0. 05), between age and local bias (r(44)=-. 19,p>. 05), nor age and global bias (r(44)=-. 29,p>. 05). In support of findings by Jarrold et al. , (2000), females showed a significantly higher mean RET score (M=28) than males (M=25) (t(42)=2. 42, p