Table 1 Chronological list of studies examining working memory in people with AN

Israel et al. (2015) [20]

BETTER

Female adults with ED:

ED-R (n = 19)

ED-BP

(n = 27)

Limitation: no control group, although the authors’ highlight that this is a pilot study.

N-back task with variable cognitive load (arithmetic) and stress (positive and negative feedback)

fMRI (1.5 T)

ED-R performed consistently better than the ED-BP group on all N-back versions. Further, the ED-R group had increased right DLPFC and premotor cortex activation during the 2-back vs. 0-back task in comparison to ED-BP. ED-BP had weaker WM activation than ED-R.

Binge ED symptoms are associated with worse WM performance whereas restricting ED symptoms are associated with better WM performance. Right posterior prefrontal cortex activation is weaker during WM in those with binge ED. Age, BMI, Education influenced these findings, whereas anti-depressant medication and chronicity of illness did not.

Weider et al. (2015) [28]

WORSE

Female adults with ED:

AN (n = 40)

BN (n = 39)

HC (n = 40)

Working Memory Index (WAIS-III Manual): Paced Auditory Serial Addition Test (PASAT) 3, PASAT 2, WAIS-III (Letter Number Sequencing,

Digit Span), WMS-R (Spatial Span)

None

The AN group had lower WM scores than both BN and HC.

Lowest lifetime BMI and depressive symptoms explained the worse WM performance in the BN group but not the AN group.

Lao-Kaim et al. (2014) [31]

NO DIFFERENCE

Female adults with ED:

AN (n = 31)

HC (n = 31)

N-back task (0, 1, 2 and 3 back). The authors’ specifically examine verbal WM, incorporating the phonological loop, the phonological store, sub-vocal rehearsal and the central executive.

fMRI (1.5 T)

No significant difference in WM task performance.

All groups showed increased activation in

the bilateral IPL, bilateral middle and superior frontal

gyri extending into the DLPFC, left precuneus

and right insula. The AN group additionally showed positive trends in the left middle temporal gyrus,

right precuneus and left IFG.

Duration of illness may be associated with lower WM accuracy in the AN group. However, anxiety and depression scores were not shown to influence WM ability in AN.

Kothari et al. (2013) [26]

BETTER

Male and female 10 yr old children of mothers with ED (n = 6192)

The counting span task - a

computer-based task that simultaneously assesses the

processing and storage components of WM.

None.

Increased WM capacity in children with mothers of AN compared to those whose mothers did not have a history of AN (e.g. those with BN or non-ED mothers).

Higher maternal education level and child IQ level may mediate the effect of better WM scores.

Brooks et al. (2012) [24]

BETTER

Female adults:

R-AN (n = 13)

HC (n = 20)

The N-back task (1-back and 2-back), presented on a computer screen, with additional subliminal images of food, neutral and aversive scenes

None

Females with R-AN were significantly better at the N-back task (fewer total errors), compared to HC. However, their superior performance on the N-back task was compromised only when subliminal images of food were presented. This suggests that subcortical (e.g. non-conscious) processing of food stimuli interfered with WM capacity.

Higher levels of anxiety correlated with number of errors during the WM task in those with R-AN.

Pruis et al. (2012) [25]

NO DIFFERENCE

Female adults:

Recovered AN (n = 15)

HC (n = 16)

Memoranda arrow was shown, followed by a distracting image of a body (negative, neutral, positive or scrambled), and then another arrow. Participants were instructed to indicate whether the second arrow presentation was in the same orientation as the first.

fMRI (3 T)

HC and recovered AN wormen did not differ on the WM task overall. Additionally, body images that were rated as negative were more disruptive to WM processes in both groups, but presentation of other distracting body images had no effect on WM. Amygdala and fusiform

activation were greater in women who

had recovered from AN than in controls

when viewing images of bodies during

the WM task. There were no group differences in DLPFC

activity. However, there was more suppression

of medial prefrontal cortex activity in women who had recovered from AN in comparison to controls when negatively rated images were presented

during the working memory task.

More years recovered may have had an influence on negative ratings of bodies and activation levels in the amygdala.

Nikendei et al. (2011) [30]

NO DIFFERENCE

Female adults:

R-AN (n = 34)

BP-AN (n = 19)

WS-AN (n = 16)

HC (n = 30)

Wechsler Memory Scale Revised (WMS-R) – Digit span backwards.

None.

Currently ill and weight-restored

AN patients did not differ significantly from healthy controls with respect to WM. However, there was impaired mmediate and delayed verbal recall performance in acute AN patients that was found irrespective of

AN subtype, and that persisted in weight-restored AN patients.

ED symptoms or comorbidities did not correlate with WM performance.

Hatch et al. (2010) [23]

BETTER

Female adolescents

AN (n = 37)

HC (n = 45)

IntegNeuro-computerized

Battery using the N-back continuous performance test of sustained attention.

None.

During underweight status, AN patients had superior WM capacity in comparison to HC.

ED symptoms or comorbidities did not correlate with WM performance.

Dickson et al. (2008) [22]

BETTER

Female adults:

R-AN(n = 24)

HC (n = 24)

The N-back task (1-back and 2-back), presented on a computer screen, with additional subliminal and supraliminal images of food, neutral and aversive scenes

None.

Participants with AN had superior WM performance compared to the HC during the subliminal condition, but were more distracted than HC by the supraliminal condition.

Duration of illness correlated positively with number of errors made by the AN group.

Fowler et al. (2006) [29]

NO DIFFERENCE

Female adults:

AN (n = 25)

HC (n = 25)

Cambridge Neuropsychological Test Automated Battery (CANTAB). Spatial WM is a test of spatial working memory

and strategy performance to find

individually hidden ‘blue tokens’

without returning to a box where

one has previously been found

None.

No impairments were observed in spatial WM.

ED symptoms or comorbidities did not correlate with WM performance.

Seed et al. (2002) [27]

WORSE

Female adults:

AN (n = 20)

HC (n = 20)

Spatial WM as part of a test battery. A picture of a house is presented for 5 s. The house has

nine windows, four of which are lit. A series of 36

presentations of the same house in which just one window

is lit follows, and the participant has to respond ‘Yes’ if

the window was one of the four lit in the original

presentation, or ‘No’ if it was not. Cortisol measures were also taken.

None.

WM performance was impaired in females with AN compared to HC, but cortisol levels did not differ between groups.

ED symptoms or comorbidities did not correlate with WM performance.