Introduction
Adolescent-onset schizophrenia (AOS) was commonly considered as a chronic and sever illness with poor prognosis and typical age of onset at 13–18 year [
1‐
4], affecting approximately one child in two hundred [
5]. Nearly a third of adult patients with schizophrenia developed their earliest symptoms in the period of children and adolescents, characterized by high diagnostic stability [
5]. According to peviousstudies, the onset of schizophrenia is thought to be basically associated with the combined effects of genetice, developmental, and environmental factors. The adolescent brain was advancing towards maturity, along with the development of the synaptogenesis and myelination [
6], suggesting that it may be easier to identify the true etiology of schizophrenia to primarily concern AOS rather than adult schizophrenia [
7‐
9].
In recent year, patients with AOS increasingly suffered with alcoholism, substance abuse and high suicide rates, which may contribute to the premature psychotis symptoms, aggravating manifestations and poor prognosis of AOS [
10]. Suicidal behaviors before the first hospitalization occurred in approximate 32% of patients with AOS, and follow-up investigations reported a suicide rate of 12% in the two years after the first admission [
11]. Urbanization and social adversity accelerated the occurrence of the harmful behaviors of AOS [
12]. Furthermore, risk behaviors were reported to be related with poor DM ability and compelling evidence has proved that DM is one of the strongest predictive factors of risk behaviors [
13,
14]. Therefore, it is of great necessity to assess the characteristics of DM in AOS.
DM is the capacity to modulate the advantageous and disadvantageous perception, facilitating rewarded consequences [
15,
16], which is generally considered as two types: one is DM under ambiguity (with uncertain probability); the other is DM under risk (with certain probability). DM under ambiguity generally means that decision makers are unaware of the outcomes and probabilities in the beginning, after which they gradually acquire favorable information and optimal choice from the feedback of previous choices [
17]. In contrast to DM under ambiguity, DM under risk is associated with explicit information and probabilities which are initailly informed [
18]. Moreover, there are representative paradigms for DM under ambiguity and under risk, i.e. IGT (Iowa Gambling Task) [
19] and GDT (Game of Dice Task) [
20], respectively. IGT, frequently revealing decisions under ambiguity and GDT usually designed to examine DM under a specific risky condition, are both well established and widely applied in the assessment of the neuropsychiatric patients [
21‐
24]. In the past decade, researchers have proceeded with numerous studies to investigate the features of DM deficits in schizophrenia [
25‐
28], which may implicate working memory deficits [
29], reward-driven hyperactivities [
30,
31], or executive dysfunctions [
32]. However, up to date, few consensuses were reached on this issue. Lee et al. (2007) firstly found adult schizophrenia has difficulty in making decisions under ambiguous condition (IGT) but not under risky condition (GDT) [
24], whereas, the results of subsequent studies tended to identify the impairments in the schizophrenia under both ambiguous and risky conditions [
25,
26,
32,
33]. Consequently, it is of importance to take the above-mentioned DM-related characteristics into account in patients with AOS, in order to reach a general agreement on the DM dysfunctions in the schizophrenia.
With the perspective of DM development, people in adolescence usually preferred to take risks and displayed insensitivity to punishment [
34]. Dana et al. (2012) found ambiguous DM abilities progressed in J-shaped curve from 8 to 17 years of age [
35], demonstrating children from 8 to10 years of age performed better in IGT than those from 10 to 12 years of age [
36], and then the performances on IGT gradually improved in those from 13 to 18 years of age. The compelling evidence indicated accessing the GDT performances may predict the gambling behaviors in adolescents [
37]. Studies on DM in healthy adolescents were considered as a foundation, accelerating researches’ investigation on the teenagers with mental or neurological diseases. However, at present, only Kester et al. (2006) investigated the characteristics of DM in patients with AOS under ambiguity using IGT [
38]. Patients with AOS exhibited worse during the last two blocks of IGT. They have trouble finding rules to avoid the disadvantageous decks from previous feedback under ambiguity. However, due to a very limited simple size (15 patients with AOS), Kester et al. (2006) may not efficiently demonstrate DM performance in AOS [
38]. Meanwhile, to our best knowledge, GDT has not yet been applied to study the characteristics of DM under risk in patients with AOS, though it is widely used in adolescent patients with neurological disease and healthy adolescents [
37,
39,
40]. Accordingly, it is of importance and interest to further investigate the performances of DM in AOS in a relatively large sample size, adopting the two different conditions (under ambiguity and under risk).
However, to date, no study has virtually compared the DM performances of AOS and healthy controls under ambiguity and risky respectively at the same time depending on a relatively large sample size, in which the evaluation of the characteristics of DM of AOS would be made more comprehensively and accurately. In the present study, we sought to prove the following hypotheses: (a) patients with AOS demonstrated DM impairments under both ambiguity and risk, utilizing experimental paradigms of IGT and GDT, respectively; Moreover, (b) the performances of DM may be associated with the psychiatric symptoms or cognitive functions in patients with AOS.
Discussion
Our findings indicated that patients with AOS showed worse performance than healthy controls in DM, especially under risky condition, as measured by IGT and GDT. In regard to the ambiguous situation, patients with AOS showed significant difference on the 5th block and failed to show learning effect on IGT. Similarly under risk, patients with AOS not only conducted more disadvantageous choices, but also exhibited worse negative feedback rate on GDT. Apart from DM impairment under risk diminished DM abilities under ambiguity were also found to be related to the poor executive function in the present study.
The DM performance in patients with AOS in our present study principally replicated the findings demonstrated by Kester et al. (2006) [
38]. Patients with AOS documented a different learning curve compared with their non-psychiatric peers. Although patients with AOS were able to explore an advantageous strategy from the feedback condition on each of the first four blocks, this strategy was incapable of developing a favorable long-term one, and progressed to a high-risk and low-reward strategy on the 5
th block. One potential factor may be patients with AOS are implicated with DM dysfunctions under risk, rather than with decisions under ambiguity. Recent evidence has suggested that people had come to learned the rules and contingencies implied in distinct options [
34,
49]; and gradually get an awareness of which decks were advantageous or disadvantageous when they performed DM on 3
rd or 4
th block during the IGT, then making decisions with certainty to the end.
In addition, compared to healthy controls, net scores of the blocks of IGT haven’t shown any difference until progressing to the 5
th deck in patients with AOS, while those have showed difference scince the 3rd or 4th deck in adult patients with schizophrenia [
26,
30]. This delay in patients with AOS was probably due to the instability in adolescence [
50], characterized by the slow development and the biological changes of the brain regions in the puberty [
50‐
52]. Adolescents are originally sensitive to immediate gains and losses (inhibitory processes), then affecting the further DM strategies on IGT [
53‐
55]. AOS showed comparatively delated DM strategy, coinciding with the adolescent development trajectory [
50,
53,
56]. Notably, the prefrontal cortex, with the properties of protracting structural development during adolescence and early adulthood, has been proved as the key brain area of DM, participating in neural processes of DM [
50,
52]. Taking into account the alterations of reward processing in the etiology of schizophrenia [
52], combining with the late developed impulse control in the adolescence [
56], performances of DM in patients with AOS might be consequently damaged. In the future, more researches with a larger sample size and different stages of adolescents with AOS were needed to verify these findings.
In GDT, patients with AOS showed remarkly deficits on risky DM. This result is consistent with the previous studies presented in adolescence patients with neuropsychosis by other scholars [
40,
57,
58]. Patients with AOS failed to efficiently integrate consequences and probabilities, and were unable to modify the options from rewards and punishments feedback, possibly lacking a vision on low risk for the higher returns. Our results found patients with AOS not only liked the risky options, but also preferred a higher number in GDT. Similarly, patients with AOS showed worse on processing negative feedback than healthy controls, rather than positive feedback. Patients with AOS were unable to transform a safe/gain option after receiving a big penalty. In other words, AOS preferred to take risks without regard to the consequences of behaviors. Another reason was that AOS may be lack of the ability of impulse control, failing to stop risking actions, and then affecting monitoring and modifying profitable dysfunctions under risk. Previous studies have always stood by the social cognition (especially DM) deficits may be qualified as a hallmark of schizophrenia. However, combining with progressive difference of DM performances occurring only on the 5
th block on IGT and high-risk preference in GDT, our results went a step further to indicate that deficits of DM under risk may be a specific socio-cognitive phenotype of AOS.
Results from the current study demonstrated lower score on IGT was related to poor performance executive function in AOS, represented by stroop test and digital span backward. Similarly, blocks and groups interactions were not significant after adding executive function as a covariance on IGT. Executive function plays an inevitable role in DM behaviors, and IGT was called as“hot”executive function in the previous studies [
59,
60]. Meanwhile, there is a positive correlation between net/total score on IGT and score on the digital span backward but not the digital span forward in the present study, which may be most likely attributed to the asumption that digital span backward required more cognitive processing, especially depending on the executive resources [
61]. These associations concurred with those shown in other previous studies [
34,
39,
62]. However, different with previous researches during the DM processes, our present study showed an inconsistent performance that clinical symptoms were unrelated to DM processes. Evidence from previous studies suggested negative symptoms were associated with abnormality of DM processes in schizophrenia through the reinforcement learning [
51,
63,
64] and similar results have been found in some fMRI studies. The most probable explanation of these inconsistent results may be that the enrolled patients with AOS in the present study were all in a short course (less than three years), and the possibly limited clinical symptoms may impede the clear presention of the relationship to a certain extent.
In together, a few limitations should be acknowledged. Although the sample size in our present study is larger than any one in the previous published studies of AOS, it is obviously not large enough to be subdivided into different age groups. Second, patients with AOS are given medication during the research. Despite they are all at a stable stage of the illness for more than three month before assessment, we were not able to completely exclude the medicinal impact. The ongoing study should recruit treatment-naive patients to clarify the mechanism of DM deficits under less interference. Third, although the healthy controls were enrolled with psychiatric interview eliminating the history of neuropsychiatric disorders, PANSS was not used to exclude the individuals with mental symptoms. Future studies could be designed including the assessment clinical symptoms.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.