International Journal of Risk and Recovery REVIEW ARTICLE Neurocognitive predictors of confabulation in Schizophre- nia: a systematic and quantitative review Kyrsten M. Grimes1, Konstantine K. Zakzanis1 1 University of Toronto Scarborough, Department of to the use of interviewing techniques. Ad- Psychology, Toronto, Canada ditionally, identifying predictors of confabu- Confabulations, or false memories, are ob- lations in schizophrenia may allow for a served in various disorders, including schizo- greater understanding of what subgroup of phrenia. In forensic psychiatric assessment, patients is more likely to confabulate. This this is problematic, particularly when garnering quantitative review will aim to synthesize a clinical history and detailed account of the the findings on the frequency of confabula- index offense(s) from the individual being tions in these studies, as well as the neu- charged. This study sought to quantitatively synthesize the existing literature regarding the rocognitive predictors of confabulation. To frequency of confabulations in schizophrenia the best of our knowledge, no quantitative and its neurocognitive correlates. The findings review employing meta-analytic methods suggest that patients with schizophrenia con- has been undertaken on the neurocogni- fabulated more than healthy controls for new tive predictors of confabulations in this information if it was related to old information. patient population to date. The relationship between confabulations and neurocognitive variables was inconsistent. Confabulations Together, the results from this quantitative Variations of the Deese–Roediger– review has important implications for interview- McDermott paradigm [8,9] tend to be used ing techniques in forensic psychiatric assess- most frequently to measure confabula- ment. Specifically, the assessor should take great care not to ask leading questions or in- tions. Participants are shown a list of troduce unverified, contextual information into words. They are then shown these words the interview, as it may result in a confabula- again, along with new words that either tion, rather than a more accurate account of are or are not semantically related to the the event. first set of words. Patients are asked to Key words identify whether they have seen the word before, as well as how confident they are Schizophrenia, confabulation, false memories, intru- in their answers. An intrusion, whereby sions, neurocognition, source monitoring participants state a new word was previ- Introduction ously presented, is considered to be evi- dence of confabulation. Research sug- Confabulations, or false memories, are gests that patients tend to confabulate for observed in various disorders, including new, semantically related words more than schizophrenia. In forensic psychiatric as- do controls [10]. In contrast, several stud- sessment, this is problematic, particularly ies have found that both healthy controls when garnering a clinical history and de- and patients made comparable levels of tailed account of the index offense(s) from intrusions when the new words were se- the individual being charged. Numerous mantically related to the old words. Fur- studies have identified that patients with ther, as the semantic relationship de- schizophrenia confabulate more frequently creased, the number of intrusions de- than do healthy controls [1-4], though oth- ceased across both groups [5,11-13]. This er studies have found no differences be- suggests that anyone is susceptible to tween these groups [5-7]. Confabulations making confabulations when the new and in schizophrenia have important implica- old words are related semantically. tions in forensic assessments with respect Int J Risk Recov 2018;1(2):23-31 https://doi.org/10.15173/ijrr.v1i2.3492 Grimes & Zakzanis IJRR 2018;1(2) A common variation to this paradigm in- Peters, Hauschildt, Moritz, and Jelinek [18] volves both the participant and experi- employed a similar methodology using menter generating words. Participants are videos but varied their valence: positive, later presented with the generated words negative, neutral, and delusional. Patients and new words and are asked to identify with schizophrenia confabulated more whether the word is old or new. If they frequently than healthy controls for posi- identify it as old, they are asked whether it tive videos only. That is, for negative, neu- was said by the experimenter or them- tral, and delusional videos, no differences selves. Patients were more likely than in confabulation were found between pa- were controls to attribute new words to tients with schizophrenia and healthy con- both the experimenter and the self [10,14] trols. This suggests that emotionality may but tended to display a bias in labelling play a role in confabulation, which makes words as being said by the experimenter sense, given that patients’ real-world con- [14]. Thus, this may simply be a source fabulations tend to have an emotional va- monitoring deficit, rather than evidence of lence. This further speaks to the multitude confabulations. Indeed, other research of factors that likely play a role in predict- suggests that patients were more likely to ing the likelihood of confabulations. misattribute self-presented words as being Neurocognition said by the experimenter and vice versa [13]. Given that confabulations are related to one’s ability to form memories, it is neces- If patients with schizophrenia confabulate sary to examine the neurocognitive predic- more than healthy individuals, this effect tors associated with this phenomenon. For should be observed using other stimuli as instance, Nienow and Docherty [20] found well. Mammarella et al. [15] asked partici- that patients with schizophrenia were more pants to either imagine an action or per- likely than controls to confabulate, but this form an action. Twenty-four hours later, effect disappeared when intellectual ability they were presented with several actions and verbal working memory was taken into and asked if they imagined it, performed it, account. Because confabulations are as- or neither (i.e. a new action). Patients in- sociated with other disorders as well, it is correctly attributed new actions as previ- possible that it is linked to specific neu- ously performed actions more than did rocognitive deficits, rather than being as- controls, suggesting that patients did in sociated with schizophrenia more broadly. fact confabulate that they had previously performed an action. Various studies have identified that both semantic and working memory differenti- Other studies have used pictures and vid- ated confabulators from non-confabulators eos, which may tap into the visual aspect but not episodic memory [5,6,21,22]. While of confabulations (i.e. being able to visual- Nienow & Docherty [20] found that verbal ly represent a memory in one’s mind). working memory was associated with con- Several studies employing the DRM para- fabulations, thought disorder accounted for digm using pictures found that like when variance above and beyond verbal work- using words, both patients and controls ing memory. Indeed, several studies have confabulated at a similar frequency when found that thought disorder is uniquely new images were highly related to previ- associated with the tendency to confabu- ously presented images. In contrast, how- late [12-14,23,24]. Thus, it appears that ever, patients confabulated more frequent- thought disorder plays a specific role in ly than did controls on new images that confabulations but that semantic and were moderately related to previously pre- working memory may be uniquely associ- sented images [16,17]. This suggests that ated with this phenomenon above and while both patients and controls have a beyond symptomatology. tendency to misattribute new stimuli as previously presented when they are relat- In general, previous findings suggest a ed, patients with the illness have a greater relationship between deficits in executive tendency than do healthy controls to do functioning and confabulations. These this when the stimuli are not related. findings have been consistent for both 24 Confabulation in Schizophrenia IJRR 2018;1(2) source monitoring paradigms [5,10,24] searching the literature was set. Hence, and story-recall [22]. On the other hand, the research literature was canvased up to other studies have yielded contradictory 2016. Studies included ranged in publica- findings [14,21]. Thus, these differences tion date from 1995 to 2007. need to be further understood. Inclusion Criteria Other neurocognitive domains have re- The search yielded 250 candidate re- ceived less attention to date than have search papers. After duplicates were re- memory and executive functioning. Atten- moved, five primary studies met inclusion tional abilities do not appear to play a role criteria. The following inclusion criteria was in confabulations [25], but this has only utilized: (1) Participant samples that in- been examined in a limited number of cluded patients with schizophrenia and studies. Moreover, Brebion et al. [25] ex- healthy controls; (2) commercially availa- amined processing speed and found this ble neuropsychological test measures to be unrelated to intrusions. It is possible were employed (i.e., no experimental par- that while processing speed did not play a adigms were considered); (3) quantitative role in the source discrimination task over- data (i.e., means and standard deviations) all, it may still have played a role in the were available so