The 5-choice continuous performance test: evidence for a translational test of vigilance for mice

Abstract

Background: Attentional dysfunction is related to functional disability in patients with neuropsychiatric disorders such as schizophrenia, bipolar disorder, and Alzheimer's disease. Indeed, sustained attention/vigilance is among the leading targets for new medications designed to improve cognition in schizophrenia. Although vigilance is assessed frequently using the continuous performance test (CPT) in humans, few tests specifically assess vigilance in rodents.

Methods: We describe the 5-choice CPT (5C-CPT), an elaboration of the 5-choice serial reaction (5CSR) task that includes non-signal trials, thus mimicking task parameters of human CPTs that use signal and non-signal events to assess vigilance. The performances of C57BL/6J and DBA/2J mice were assessed in the 5C-CPT to determine whether this task could differentiate between strains. C57BL/6J mice were also trained in the 5CSR task and a simple reaction-time (RT) task involving only one choice (1CRT task). We hypothesized that: 1) C57BL/6J performance would be superior to DBA/2J mice in the 5C-CPT as measured by the sensitivity index measure from signal detection theory; 2) a vigilance decrement would be observed in both strains; and 3) RTs would increase across tasks with increased attentional load (1CRT task<5CSR task<5C-CPT).

Conclusions: C57BL/6J mice exhibited superior SI levels compared to DBA/2J mice, but with no difference in accuracy. A vigilance decrement was observed in both strains, which was more pronounced in DBA/2J mice and unaffected by response bias. Finally, we observed increased RTs with increased attentional load, such that 1CRT task<5CSR task<5C-CPT, consistent with human performance in simple RT, choice RT, and CPT tasks. Thus we have demonstrated construct validity for the 5C-CPT as a measure of vigilance that is analogous to human CPT studies.

Figure 1. Schematic of the 5C-CPT stimuli.

Example of the two trial types in the 5C-CPT. Go trials (relevant stimuli) appear 83% of the time, and the mouse must respond to the stimulus by nose-poking beyond the infra-red (IR) beam in the location of the cue stimulus. Cue stimuli can appear in any one of the five locations. No/go trials (irrelevant stimuli) occur 17% of the time, all five cue lights come on, and the mouse must inhibit from responding in any of the five locations.

Figure 2. Strain performance at baseline in the 5C-CPT.

Performances of C57BL/6J and DBA/2J mice were compared at baseline in the 5-choice continuous performance test (5C-CPT). The 5C-CPT was adapted from the 5CSR task and includes no-go trials, requiring greater stimulus and inhibitory control. The two strains of mice demonstrated equal ability in discriminating between target locations as measured by accuracy (A). A trend towards greater %Omissions in DBA/2J mice was observed however (B). No significant strain effects on mean correct latency (C) or premature responses (D) were observed. The inclusion of no-go trials in the 5C-CPT allowed the use of signal detection theory, with which the sensitivity index (SI) could be calculated as an index of vigilance performance, consistent with human CPTs. Differences in SI were observed between the two strains (E), although this was not significant. SDT was also used to calculate responses index (RI) bias (F). There were trends toward poorer SI levels and more conservative responding in DBA/2J mice. Data presented as mean+s.e.m., and $ denotes p<0.1 when compared to C57BL/6J mice.

Figure 3. Strain performances in the 5C-CPT extended session.

5C-CPT performances of C57BL/6J and DBA/2J mice were compared in an extended session challenge (250 trials), to increase attentional load. Consistent with baseline performance, the two strains of mice demonstrated equal ability in discriminating between target locations as measured by accuracy (A). DBA/2J mice exhibited significantly higher levels of %Omissions (B), and a slower mean correct latency (C), compared to C57BL/6J mice however. The challenge did not result in differences in premature responses (D), but significant differences in vigilance performance SI (E) was observed between the two strains, with poorer performance in DBA/2J mice. The DBA/2J mice also exhibited a significantly more conservative response bias compared to C57BL/6J mice, based on responsivity index (RI; H). Data presented as mean+s.e.m., and * denotes p<0.05 when compared to C57BL/6J mice.

Figure 4. Mouse vigilance decrement in the 5C-CPT over time.

The performances of C57BL/6J and DBA/2J mice over time in the 5C-CPT were compared. Performance was binned into 50 trial blocks to ensure consistency in trial number across blocks so that proportional data could be compared. No strain by trial block interaction was observed for any measure. A significant main effect of trial block was observed for SI (A) however, indicative of poorer vigilance with cognitive fatigue. No effect of trial block was observed for responsivity index RI (B), indicating that the deterioration of cognitive performance over time was not confounded by physical factors, consistent with human CPT performance. Data presented as mean+s.e.m., and * denotes p<0.05 when compared to trial block 1.

Figure 5. 5C-CPT extended session performance of mice binned by ITI time.

The 5C-CPT included a variable inter-trial interval (ITI; variable period after which the cue stimulus can appear), ranging from 3–7 s. Performance of the two strains was binned according to ITI time and compared, to assess the effects of ITI bin on performance. No interaction between strain and ITI time was observed for any measure. Increased premature responses were observed with increasing ITI time (A), demonstrating the temporal impulsivity resulting from consistent training in this variable ITI. No significant effects of ITI time on accuracy (B) were observed, nor on %Omissions (C). A significant main effect of ITI time on RT was observed, with performance being fastest at the middle ITI times. No significant main effects of ITI time on SI (F) were observed, although, consistent with %Omissions and RT, the best performance appeared at the center ITI times. Data presented as mean+s.e.m., and * denotes p<0.05 when compared to ITI time indicated.

Figure 6. Comparison of percentage non-responding in no-go vs. go trials in the 5C-CPT.

Given the similarity in response type for a correct rejection in a no-go trial to a miss in a go trial (both are represented by a lack of response), it is important to establish that the performance of the mice is dependent upon trial type. C57BL/6J and DBA/2J mice exhibited comparable levels of %Correct Rejection (%Corr Rej) during no-go trials in the extended session 5C-CPT. During go trials however, C57BL/6J mice exhibit significantly lower levels of %Omissions compared to DBA/2J mice. Importantly, both groups exhibited significantly different levels of non-response to each trial type, suggesting that they performed differently during go trials compared to no-go trials, and hence they their response was dependent upon trial type. Data presented as mean+s.e.m., and * denotes p<0.05.

Figure 7. Performance of C57BL/6J mice across simple and choice RT tasks as well as in the 5C-CPT at baseline.

Three separate groups of C57BL/6J mice were trained in a simple (1CRT task) and choice (5CSR task) RT tasks as well as the 5C-CPT. Significant differences in performance were observed across tasks, where consistent with human performance, RT was fastest in a simple RT task, and slowest in a vigilance task (A). Performance also differed as measured by premature responses, with the greatest levels observed in the simple RT task requiring the lowest level of inhibitory control (B). Increased %Omissions were observed with increased attentional load, although these effects were not significant (C). Data presented as mean+s.e.m., * denotes p<0.05 and $ denotes p<0.1 when compared to tasks indicated.

Figure 8. Performance of C57BL/6J mice across simple and choice RT tasks as well as in the 5C-CPT in the extended session.

The performance of three separate groups of C57BL/6J mice was compared with an extended session challenge, in simple (1CRT task) and choice (5CSR task) RT tasks, as well as the 5C-CPT. Significant RT differences were observed across tasks, where consistent with human performance, RT was fastest in a simple RT task, intermediate in the choice RT task (5CSR task), and slowest in a vigilance task (5C-CPT; A). A significant difference in premature responses, with the greatest levels being observed in the simple RT task that required the lowest level of inhibitory control (B). Increased %Omissions were observed with increased attentional load, although these effects were not significant (C). Data presented as mean+s.e.m., * denotes p<0.05 and $ denotes p<0.1 between tasks indicated.