The neurocognitive functioning in bipolar disorder: a systematic review of data

It has been reported that both patients with BD and their families have above average IQ and general intellectual functioning [23‐29], or at least they have intelligence similar to healthy controls [30‐34]. On the other hand, higher functioning and preserved neurocognitive performance during the premorbid phase, as well as higher social class could push towards a mood disorders diagnosis rather than schizophrenia [25, 30, 34‐45].

Overall, the literature reports that patients with BD manifest moderate global reduction in their neurocognitive functioning as reflected in their IQ scores and their performance in neuropsychological batteries, irrespective of illness phase [43, 44, 46‐48]. It has also been shown that these deficits are milder but may be qualitatively similar to those seen in patients with schizophrenia [21, 49]. The impairment is more severe in the presence of psychotic features [50], and, in line with this, it has been suggested that the deficit is comparable to that seen in schizophrenia, especially during the acute psychotic manic phase [51‐54].

One study suggested the presence of a more complex relationship between BD and IQ since it has been reported that although BD was related to higher premorbid IQ, further analysis revealed that men with the lowest and the highest IQ (especially verbal or practical ability) were at the greatest risk for pure BD (without comorbidity) [55]. There are inconclusive data concerning whether psychotic BD is related to impaired premorbid IQ [56, 57].

It is well known that the full IQ is composed of two composite scores; the Verbal IQ (VIQ) and the non-verbal or Performance IQ (PIQ). The effect size of the deficit during the manic phase is 0.06 and 0.28 for PIQ and VIQ, respectively [53]. It seems that during acute mania the effect size is equal to 0.47 for the full IQ [54]. There are no studies concerning the IQ in patients during the acute depressive phase. Since it is possible, the premorbid IQ of BD patients is higher in comparison to the normal population, the true magnitude of decline cannot be evaluated accurately, and all studies reporting this decline in comparison to population norms underestimate it.

According to a meta-analysis, patients with BD show higher VIQ in comparison to PIQ scores [58, 59]. This is due to the uniform reduction in all PIQ subtest scores with an accompanying preservation of VIQ scores. This impairment is not because of the slowing in mental speed which is observed in BD patients (since four PIQ subtests in comparison to one VIQ subtest are timed) and this is obvious from the quality of the deficit which persists during periods of euthymia, and it is confirmed from targeted research [60]. The deficit might be already present during the early stages of BD [61] although more recent data argue against this [62]. One study has reported that the severity of depression seems to reduce VIQ scores and thus it might diminish the VIQ-PIQ discrepancy [63]. It is important to notice that this discrepancy does not seem to be present in patients with unipolar depression [64].

It has been reported that the VIQ-PIQ discrepancy might reflect a specific effect of BD on ‘fluid intelligence’ (the capacity to think logically and solve problems in novel situations, independently of acquired knowledge) with a simultaneous respect of the ‘crystallized intelligence’ (the ability to use skills, knowledge, and experience; it does rely on long-term memory). However, any attempt for a deep understanding of this deficit is risky and premature [63].

When all phases of the illness are taken into consideration, the effect sizes concerning current IQ reduction range from 0.36 to 070 [63, 65]. For patients in remission, the results of meta-analyses are inconclusive. The effect sizes reported for PIQ range from 0.40 to 0.50 [63, 66] to lower and within the normal range, that is 0.11–0.16 [8, 12]. A problem is that most meta-analyses report an effect size for the full IQ and not separately for VIQ and PIQ. The effect size of premorbid IQ change in euthymic BD patients is reported to be low and not significant (0.04–0.20) [66, 67].

Although mental speed and psychomotor activation are two concepts which overlap and include reaction time, cognitive and motor speed and, manual dexterity, they are clearly not identical. Additionally, most of the neuropsychological tools which are used for the evaluation of psychomotor and mental speed also assess other neurocognitive functions, and this is at least partially a consequence of a methodology effect, because to measure ‘speed’, you need to initiate a ‘procedure’ whose ‘speed’ is going to be measured.

It has been reported that the reaction time in bipolar depressive patients is prolonged [68]. Also, euthymic patients had not only prolonged reaction times but also more error rates in a visual backward masking test. Overall, reaction time prolongation has been found to be associated with burden of illness and especially past history of depressions but not with current medication [69].

Irrespective of illness phase and symptom severity, the speed of mental processing in BD patients seems to be slower [34, 43‐45, 59, 70‐87]. Moreover, it has been reported that the speed of mental processing is less affected in comparison to schizophrenia [38, 40, 61, 88], although there are some data suggesting the presence of a similar degree of impairment between patients with BD and patients with schizophrenia [89]. On the other hand, it has been suggested that there were no significant differences between depressed BD patients, euthymic BD patients and healthy controls in psychomotor speed [90]. Another study suggested that patients with BD showed a linear improvement in processing speed in the first year after resolution of their initial manic episode [91]. It has been reported that bipolar depressives manifest slower mental speed in comparison to manic and unipolar depressives, even after corrected for motor speed. The interesting feature in that particular study was that distraction improved the mental speed of BD depressed patients while it adversely influenced the speed of the other two groups [92].

The deficit could be present already during the early stages of BD [93]. Individual studies suggested that the magnitude of mental speed impairment in patients with BD is reported to correspond to an effect size of 0.82–1.08 [48, 77, 94]. When all phases of the illness are taken into consideration, the effect size 0.50–0.55 (which is similar to that observed concerning the IQ) [63, 65]. In euthymic BD patients, one meta-analysis reported that the effect size with the use of the TMT-A was 0.52 and with the DSST was 0.59 [13]. A second study reported effect sizes of 0.60 and 0.79, respectively [67], a third one reported effect sizes equal to 0.64 and 0.76 [12], a fourth 0.71 and 0.84 [8] and a fifth 0.7 and 07–0.8, respectively [66].

It is important to note that the deficit in the processing speed might have a significant confounding effect on the performance in almost all neurocognitive testing and controlling for it might make any difference between groups concerning other neurocognitive domains disappear [95]. Also, it is suggested that global functional impairment is strongly associated with poor performance on a cognitive measure of processing speed (e.g., WAIS Digit Symbol or the TMT) [75, 96].

Attention is a concept that includes a number of processes which work together, influence one another or prerequisite one another. These processes are: working memory (which refers to the ability to keep a limited number of mental objects in awareness for a limited duration of time), vigilance (which is the capacity to identify a specific target among many other stimuli), freedom from distraction or interference and the ability to split or to rapidly shift attention. Concentration is a term which refers to the ability to sustain attention over prolonged periods of time. There are many tests, with each of them assessing one of the previously mentioned processes. For example, the Continuous Performance Test assesses vigilance while ‘span tasks’ assess working memory. However, all these tests except from the specific aspect of attention they assess are also influenced from the other processes which are related to attention as well. Working memory is often classified as belonging to the executive functions and it is often considered in relation to them.

The magnitude of the attentional impairment is independent of current symptomatology and of the phase of the illness; however, significant variability is present in the literature [34, 44, 48, 71, 85, 87, 91, 97‐113]. It has been also reported that the impairment is present already during the early stages of the disorder [93] and it is less pronounced in comparison to the deficit seen in patients with schizophrenia [38, 114, 115]. On the other hand, however, some studies reported that there is a similar magnitude of impairment between patient with BD and patients with schizophrenia [89, 116, 117]. It has been suggested that the performance in divided attention (DA) varied considerable over time within patients. It was also found a significant quadratic relationship between manic symptoms and DA performance, even after corrected for the effect of psychotropic medication. It has been suggested that mild hypomanic symptoms have a positive influence on divided attention scores and moderate to severe manic symptoms have a negative influence. No association between depressive symptoms and DA performance was found [118]. The magnitude of effect sizes ranges from 0.36 to 0.82 [48, 94, 119] depending on the domain assessed and the composition of the study sample.

The overall effect size calculated after a meta-analysis is 0.64 which is similar to that reported concerning the rest of neurocognitive functions [63]. Another meta-analysis reported that attention is impaired during the acute manic/mixed state, with effect sizes ranging from 0.79 to 0.90, during the acute depressed state with an effect size up to 0.80, but also during euthymia with effect size from 0.41 to 0.65 [11]. Another meta-analysis concerning BD patients during euthymia has reported medium effect sizes (0.48–0.60 depending on the testing condition) [13], while a third one found effect sizes of similar magnitude (0.62 and 0.74 for CPT hits and reaction time, respectively) [67]. A fourth meta-analysis gave effect size 0.58 for CPT and 0.37 for digits forward [8] and a fifth reported an effect size equal to 0.8 for CPT [66].

There are also a limited number of studies which indicate no impairment in attention [79, 120‐123], and this concerns especially euthymic patients [124]. One study has found that depressed BD patients, euthymic BD patients and healthy controls had no significant differences in attention [90].

Learning refers to the ability to acquire and store new information. Memory is the mental process that allows individuals to retrieve the new information at a later time. Learning and memory involve a number of processes including attention and concentration, encoding and allocation of effort. These processes are distinct from one another but interrelated and interdependent. Moreover, there are different strategies and processes involved, depending on whether a short- or a long-term effect is desirable and also depending on the quality and nature of the information and the frame it is presented in.

A typical classification of learning and memory structure is shown in Table 2 [125]. Due to the fact that much of research on memory is focused on ‘depression’ and does not distinguish between unipolar and bipolar depression, the results and the conclusions from these studies should be received with reservation because it is uncertain whether they apply specifically to BD, and to which extent.

It has been suggested that there is a deficit in working memory [45, 57, 72, 73, 77, 84‐87, 126‐129] and specifically in the visuospatial working memory [33, 130‐134]. Psychotic patients might have worse performance [48, 135]. The impairment is probably present already since the early stages of the illness [136]. Some studies have shown that the impairment in working memory affects only patients with acute mania [53, 131, 132]. Comparing to patients with schizophrenia, the impairment in visuospatial working memory is less pronounced [115].

Additionally, there are studies indicating a deficit in declarative memory [137] and specifically in the semantic [138] and the episodic memory [48, 109, 139, 140], verbal learning [45, 47, 72, 86, 141‐147] and this is true both for BD-I and BD-II [134, 148]. A deficit is also present concerning verbal memory [38, 41, 73, 75, 87, 102, 111, 143, 144, 147, 149‐151], also during periods of euthymia [152]. One study, however, has shown that depressed BD patients showed greater impairments in verbal memory than the euthymic BD patients [90]. It has been suggested that the magnitude of the effect size of the verbal memory deficit is 0.7–0.9 [41, 139] and it is less pronounced in comparison to that seen in schizophrenia [153, 154] and cannot be explained by the attentional deficit [120]. Patients at early stages of BD have better performance in the total immediate free recall and in delayed free recall compared to patients at a later stage and to patients with schizophrenia. Additionally, concerning the ability to retain words learned, BD patients at a later stage and chronic patients with schizophrenia were more impaired than BD patients at early stage and patients with recent onset schizophrenia [155]. It has been also indicated that delayed free recall is worse in patients with BD compared to healthy controls [156].

There are impairments in total learning as well as short- and long-delay verbal recall, recognition, discriminability and learning slope [157], associative learning [158], implicit motor learning [159], immediate memory [134], delayed memory [34, 45, 72, 77, 84, 85, 94, 108, 160], non-verbal memory [161], visual memory [38, 43, 44, 74, 79, 102, 108, 111], autobiographical [162, 163] and prospective memory [164]. It has been shown that prospective memory deficits [165] and short-term non-affective memory [166] are also present in remitted BD patients suggesting that they constitute a trait deficit. Moreover, it has been shown that BD patients manifest a deficit in incidental contextual memory in the absence of a binding cue at encoding. There was no difference found between the groups for contextual memory even under incidental encoding with the binding cue. One study has indicated that the impairment in the contextual memory was reduced by providing cognitive support at encoding [167].

There are also negative studies concerning the presence of impairment in working memory [168‐170], spatial working memory [171] and in verbal learning [138] and verbal [123, 172, 173] and visual memory [173]. One study has shown that the ability to learn is maintained both by BD patients and by patients with schizophrenia [155] and another study has reported that there were no differences between BD patients and healthy controls in terms of their slope of learning, retrieval index, retention percentage, semantic or serial clustering, errors, or level of retrieval [174]. It seems that most memory impairments are due to the presence of confounding variables except maybe for verbal recall [101]. The possibility that difficulties in the semantic clustering or other strategic processing deficits are the cause for the verbal memory impairment has both positive [139, 160] and negative data [137, 139, 161, 175].

In meta-analytic studies, when all phases of the illness are taken into consideration, the magnitude of the effect size is 0.60 for working memory, 0.43 for immediate verbal memory and 0.34 for delayed, 0.26 for immediate visual memory and 0.51 for delayed [65]. The meta-analysis of short-term memory studies revealed an effect size of 0.58 when span tasks were utilized, without any difference between auditory or visual tasks. When verbal learning tasks were used, the effect size was 0.91 [63].

During the acute manic/mixed state, the magnitude of the effect sizes was 1.43 for verbal learning and 1.05 for delayed free verbal recall. Additionally, concerning the acute depressed state, the effect size for verbal memory was 1.20, while during euthymia the domains impaired were working memory (0.65), verbal learning (0.81), long-delay verbal free recall (0.78), immediate non-verbal memory (0.73) and delayed non-verbal recall (0.80) [11].

The average effect size for episodic memory in euthymic BD patients is reported to be equal to 0.62 and for working memory equal to 0.60 [12]. Again in euthymic BD patients, a large effect size (0.90) is reported for verbal learning and working memory (0.98) and somewhat lower effect sizes for aspects of immediate (0.73) and delayed (0.71) verbal memory [13]. A third meta-analysis reported effect sizes of 0.81, 0.54, 0.74 and 0.72, respectively [67]. Another meta-analysis reported an effect size of 1.02 for working memory. 0.85 for delayed recall, 0.82 for immediate recall and 0.62 for visual memory [8]. Finally, a last meta-analysis gave an effect size of 0.85 for verbal learning, and 0.73 for verbal memory-early recall [9, 66]. It seems there is a publication bias especially concerning verbal learning and after correction for this, the effect size is attenuated (from 0.85 down to 0.66) [9].

Overall, the literature suggests that BD, irrespective of illness phase, is characterized by a severe deficit in the acquisition of new information, but not in the retention [41, 63, 161, 175, 176]. In spite of some opposing data, the most probable interpretation which derives from empirical studies is that the attention and concentration deficits impair the acquisition of information and learning, by disrupting the engagement of effortful processing which results in a shallow rather than deeper level of processing (e.g., acoustic rather than semantic) [137, 175, 177‐186].

The evaluation of verbal skills includes mostly the evaluation of verbal fluency. Although the literature has reported that verbal skills are impaired during all phases of BD [84, 87, 108, 129, 147, 149, 166, 169, 187‐189], there are some studies reporting that this impairment is not present during euthymia [41, 52] or during the first mood episode [121]. One study has indicated that there are errors in speech during the acute manic state, and these errors are independent from the co-existence of an attentional deficit [106].

Concerning the magnitude of the effect size, it has been reported that it is small [169] and even smaller in comparison to the effect size that seen in schizophrenia [61], however, patients with psychotic BD have a higher effect size (0.68–1.73) [94, 117, 190]. Even smaller (below 0.50) effect sizes have been reported [63], while when all phases of the illness are taken into consideration, an effect size equal to 0.63 emerges [65]. Both letter fluency and semantic fluency are impaired during the acute manic/mixed state and the effect sizes are equal to 0.51 and 0.59, respectively. The phonemic fluency is impaired during the acute depressed state with an effect size equal to 0.93 while during euthymia impaired are both phonemic fluency (0.51), and semantic verbal fluency (0.75) [11]. In euthymic BD patients, the average effect size for verbal fluency ranges between 0.56 and 0.90 [12, 67] but it is smaller (0.34) for verbal fluency by letter [13].

The evaluation of visuospatial skills is usually made with the use of the complex Rey figure or with the WAIS-R block design. Patients with BD and their unaffected relatives show impairment in the visuospatial/constructional abilities [45, 147, 191] and in visual learning and memory [191, 192]. It is interesting that euthymic BD patients and patients with schizophrenia have similar impairment when the results are controlled for possible confounding factors [193]. Also, one study has shown that BD patients were significantly impaired on all three object location memory processes (positional memory, object-location binding, and a combined process), with the largest effect found in exact positional memory (d = 1.18) [194]. Unaffected relatives demonstrated an intermediate level of performance in comparison to patients with BD and to normal controls [195]. Contrawise, one study suggested that the visual motion integration is intact in BD patients [196]. Some authors suggest that the impairment in visuospatial skills is restricted in the acute phase while these skills might not be affected during remission [41, 63]. One study has reported that the overall effect size is equal to 0.65 [63] and other studies have reported that in remitted patients the effect size is 0.22–0.57 [8, 12].

The executive system is considered to be involved in the planning, the decision-making, the error correction and the troubleshooting, in situations where responses are not well rehearsed or contain novel sequences of actions, are dangerous or constitute technically difficult situations or situations requiring the overcoming of a strong habitual response or resisting temptation. In other words, ‘Controlling of mental and neurocognitive processes’ seems to be the key phrase describing the role of executive functions. In patients with BD, reasoning should be considered separately from the rest executive functions due to the fact that it seems to rely heavily on verbal and linguistic skills [63].

It has been suggested a severe impairment in executive functions except reasoning during all phases of BD [34, 36, 43‐45, 48, 70, 72, 73, 75‐77, 86, 87, 90, 94, 101‐103, 108, 109, 111, 120‐122, 129, 143, 146, 149, 190, 195, 197‐209] and early during the course of BD [93], but the deficit is less pronounced in comparison to schizophrenia [38, 40, 61, 89, 116, 122, 210]. However, at least a subgroup of patients is as severely impaired as patients with schizophrenia [153, 211, 212]. One study indicates that the depressed BD patients showed greater impairments in executive functions comparing to the euthymic BD patients [90]. It is suggested that this impairment might be particularly severe concerning interference and inhibitory control [47, 80, 82, 108, 168, 190, 192, 213‐215]. Additionally, patients with BD might have more risky [216] or erratic choices [217, 218] especially when a history of alcohol abuse exists [219].

However, not all data are straightforward. Normal overall executive function has been suggested by one study [144], while another reported only prolonged time to complete the test [207]. Also, a bimodal distribution of the Wisconsin Card Sorting Test (WCST) scores in patients with BD is reported, with some patients at near-control levels and others significantly impaired [41]. According to another study, the performance on the executive function measures is bimodal among euthymic BD patients (one subgroup with relatively normal and one subgroup with impaired executive functioning) [41]. Some authors argue that there is no difference between BD patients and controls in executive functions [78, 123, 170, 188, 220, 221], while others argue that the deficit is present only in the more severe and chronic cases [188]. It has been also indicated that BD patients show an improvement in the executive functions in the first year after resolution of their initial manic episode [91].

The deficit in executive functions might be overall independent from illness phase, however, there are data suggesting that some aspects of the deficit are related to affective lability [222], duration of illness, residual mood symptoms and current antipsychotic treatment [197] and history of psychosis [48]. One study found that the impairment in executive function was related to the severity of general psychopathology as it is measured by the PANSS [198] and another one correlated impaired insight with impaired executive functions [223].

A meta-analysis suggested that when taking all phases of the illness together, the effect size of this impairment is equal to 0.79 when reasoning (which was reported to be intact) is excluded [63]. Moreover, another meta-analysis reported an effect size of 0.34 for concept formation and of 0.55 for executive control, with no significant effect for current clinical condition [65]. During the acute manic/mixed state, the general executive function and the speeded set-shifting are impaired with the magnitude of effect sizes to be equal to 0.72 and to 0.64, respectively. During the acute depressed state, speeded set-shifting is impaired (0.64) while during euthymia problem-solving tasks (0.54), set-switching tasks (0.73) and verbal interference (0.75) are impaired [11]. Three recent meta-analytic studies in euthymic patients reported that the effect sizes were for executive functioning equal to 0.80 for the TMT-B, 0.56 for the WCST and 0.80 for the Stroop test [12] according to the first one, 0.78, 0.62 and 0.63, respectively, according to the second one [13] and 0.55, 0.69 and 0.71, respectively, according to a third meta-analysis [67]. Another meta-analysis reported an effect size of 0.99 for the TMT-B and 0.88 for the WCST perseverative errors and 0.52 for the WCST categories, 0.73 for the Stroop time and 0.65 for the stroop-correct [8]. Similarly, another study reported 0.70 for the WCST perseveration score and 0.8 for the TMT-B and the stroop test [9, 66].

The term ‘social cognition’ constitutes a psychological domain with several dimensions. It refers not only to the ability of the person to assume that other people have minds similar to his/her own and to interpret, but also to understand and predict the emotions, desires, intentions, behaviors and speech of others (including non-verbal elements). Social cognition shapes communication and interaction with others and in this way enabling adaptive social adaptation. It involves a complex set of processes including the representation of internal somatic states, knowledge about the self, perception of others, and interpersonal motivations [224].

The broad theory of mind (ToM) includes three main processes (a narrow definition of ToM, emotion processing, and affective decision-making). The narrow definition of ToM (mentalizing or mindreading) refers to the ability to attribute mental states (e.g., beliefs, desires, and intents) to oneself and to others. Emotion processing is the ability to identify and discriminate basic emotions. Affective decision-making is crucial for an appropriate social behavior, and concerns weighing up choices in association with reward and punishment [225].

The tests which are used to evaluate these domains are both verbal (scenarios) and non-verbal (pictures). They demand the subject to identify and comprehend the situation, the roles and the interactions and to make appropriate planning. So far, empirical data have confirmed the universality of facial emotions. This means that the specific ability to process and identify facial emotions is a substantial feature of human communication and social interaction, which is independent of culture. It has been found, even in “ecological” tests that mimic real life scenarios that patients with BD showed less impairment on neurocognitive performance compared to patients with schizophrenia [154].

Neuropsychological dysfunction in BD may also be related to the clinical symptom pattern and severity, and it has been correlated with age, earlier age at onset, medication status, as well as with idiosyncratic factors affecting the long-term course.

In spite of the research efforts during the last few decades, there has not been found a specific neurocognitive profile for the different bipolar subtypes [340]. This is probably because research on BD-II patients is rare, and there are large methodological differences between studies. As a result, inconclusive data are in place. Research on the other subtypes of the bipolar spectrum is essentially lacking. There is only one study reporting data on ‘bipolar spectrum’ patients. That paper suggested the presence of a broad neurocognitive impairment, particularly affecting verbal memory and the executive functions [341].

Some unsystematic reports in the literature suggest that BD-II patients have similar performance to controls [250, 342, 343], but others supported the notion that BD-II patients perform in-between healthy controls and BD-I patients [72, 73, 75, 79, 250, 344‐346], but this maybe specific concerning verbal memory [149, 346] and executive functions [346]. On the contrary, there are data suggesting that BD-II patients perform similar to BD-I [72, 75, 138, 160], or even perform worse than BD-I patients, at least in some specific neurocognitive domains, including reaction time and inhibition [347, 348].

The literature cannot answer the question whether there is any qualitative difference between bipolar subtypes. Such a difference would suggest (although that would not be mandatory) that possibly there are different neurobiological mechanisms which underlie BD subtypes. However, both questions (concerning a quantitative and a qualitative difference) remain unanswered. For both there are data in favor and against.

A global neurocognitive impairment might be present in BD-II patients, with only phonemic verbal fluency being preserved and with moderate to strong effect sizes ranging between 0.62 and 1.34 [345]. For premorbid IQ, there is only one study which have not found any differences between BD-I and BD-II patients, since both groups had worse performance in comparison to the performance of controls [72]. It has been also shown that the intellectual decline was less pronounced in patients with BD-II in comparison to those with BD-I [347, 349]. There are several studies suggesting an impairment in psychomotor speed and in attention [72, 73, 287, 345, 346, 348]; however, other studies did not support this [342, 344]. One study suggested that reaction time was similar to that of controls [350]. Moreover, another study showed that attention was intact but psychomotor speed was impaired [79]. As far as memory, there seems to be a deficit in verbal memory and verbal learning [79, 149, 345‐347] but some studies disagree [72, 73, 342, 344, 348]. Other authors found that in BD-II patients there is a presence of a less disorganized semantic system in comparison to patients with BD-I [138, 160]. Additionally, delayed memory is rather intact [72]. There are controversial data concerning the deficit in visual memory, since some authors reported an impairment [72, 345, 347] while others disagreed [73, 287, 342, 343, 348]. Similarly, some authors supported the presence of impairment in executive functions and in working memory [72, 79, 149, 344‐348] while others did not [342, 343]. Additionally, there is one study which showed that there is a deficit only in working memory but not in executive functions [73]. It has been observed that all the studies using the stroop color–word test, which assesses interference, showed a deficit in inhibitory control in BD-II patients [344‐347]. Also, there might be a deficit in the emotional processing domain [287, 347]; however, the emotion recognition seems to be intact [250]. It has been also found that unmedicated depressed BD-II patients had intact decision-making performance [343]. On the other hand, one study suggested that medicated BD-II patients had worse performance in comparison to unmedicated BD-II patients in sustained attention [287].

One meta-analyses suggested that there is no difference between BD-II and healthy controls neither in the estimated current intelligence quotient (IQ) nor the premorbid IQ [340, 351]. Another meta-analyses found that neurocognitive impairment in BD-II patients is as severe as in BD-I patients except for the domains of semantic fluency and memory [10]. There are contradictory data concerning psychomotor speed, verbal and visual memory and the impairment in these domains is probably small in magnitude. On the contrary, there are robust data concerning the presence of a working memory deficit and a decrease of cognitive flexibility and impaired inhibitory control in BD-II. Patients with BD-II manifest a deficit also in recognizing emotions [351]. In quantitative terms, BD-II does not differ much from BD-I [119] although it seems that the opinion which prevails is that BD-II patients manifest better performance in comparison to BD-I but worse than healthy controls and are positioned in between these two groups.

Although many patients with BD frequently complain about neurocognitive problems in attention, concentration and memory, there are limited data on the relationship between subjective cognitive complaints with objective neuropsychological deficits.

The patients’ subjective cognitive complaints do not seem to correlate or predict objective neuropsychological deficits [142, 164, 374]. One study has suggested that there might be a weak correlation between subjective complains and deficits in attention, memory and execute function [375]. Moreover, it has been shown that neither mood symptoms nor the severity of mania or depression correlate with patients’ self-report [142] One study reported that patients with BD might show more subjective complaints when there is a higher number of episodes, and especially a higher number of mixed episodes, a longer duration of the illness and when the onset of the illness occurred at an earlier age [375]. There seems to be some association between depressive symptoms and self-report neurocognitive complaints; however, the association between complaints and objective neurocognitive functioning is not moderated by mood symptoms [374].

There is limited literature concerning the role that gender might play and no conclusions can be derived. Males with BD manifest a more severe deficit in immediate memory [376] and executive function [126] whilst one study suggested that gender does not play any role on emotion recognition [250]. A better memory performance might be related with female gender [175], while another study suggested that the verbal memory deficit might relate to female gender as an endophenotype [377]. Overall, the data are conflicting with one out of three recent meta-analysis being in favor of a gender effect [8], while the two others were against [12, 66].

Endophenotype constitutes a stable pattern of behavioral symptoms with a clear genetic connection. It can be used to bridge the gap between high-level symptom presentation and low-level genetic variability. Ideally, an endophenotype might be useful in the differential diagnosis between disorders when they present with similar symptoms. Additionally, an endophenotype is associated with a specific disorder in the population, and should be heritable and state-independent. It is expected that endophenotype and the specific diagnosis co-segregate within families and the endophenotype is found in non-affected family members at a higher rate than in the general population.

The neurocognitive impairment in patients with BD is at least partially different in comparison to the deficit seen in patients with schizophrenia and it is also at least partially state-independent and definitely present during periods of complete recovery and euthymia. Although the literature so far does not support the use of global measures of neurocognition as endophenotypes [378‐380]. On the other hand, however, specific neurocognitive deficits might be relevant, although further research is needed. It has been reported that inaffective relatives of patients occupy an intermediate position between patients and healthy controls concerning specific neurocognitive functions [195].