Touchscreen and Translational Cognition: A Systematic Review of Trials in Humans and Rodents

(Full-text capture 2026-09-21; web artifacts lightly stripped; truncated.)

Tamires Coelho Martins

Tamires Coelho Martins

1Center for Technology in Molecular Medicine (CTMM), Federal University of Minas Gerais (UFMG), Belo Horizonte, Minas Gerais, Brazil

1, Renata Maria Silva Santos

Renata Maria Silva Santos

2Post‐graduate Program in Molecular Medicine, Federal University of Minas Gerais (UFMG), Belo Horizonte, Minas Gerais, Brazil

2, Rayany Karolyny da Silva Andrade

Rayany Karolyny da Silva Andrade

2Post‐graduate Program in Molecular Medicine, Federal University of Minas Gerais (UFMG), Belo Horizonte, Minas Gerais, Brazil

2, André Soares da Silva

André Soares da Silva

1Center for Technology in Molecular Medicine (CTMM), Federal University of Minas Gerais (UFMG), Belo Horizonte, Minas Gerais, Brazil

1, Felipe Baptista Brunheroto

Felipe Baptista Brunheroto

1Center for Technology in Molecular Medicine (CTMM), Federal University of Minas Gerais (UFMG), Belo Horizonte, Minas Gerais, Brazil

1, Isabella Paula Gomes Rocha

Isabella Paula Gomes Rocha

1Center for Technology in Molecular Medicine (CTMM), Federal University of Minas Gerais (UFMG), Belo Horizonte, Minas Gerais, Brazil

1, Vitória Carrazza Gambogi Loureiro

Vitória Carrazza Gambogi Loureiro

1Center for Technology in Molecular Medicine (CTMM), Federal University of Minas Gerais (UFMG), Belo Horizonte, Minas Gerais, Brazil

1, Yuri Cristelli de Sousa Silva

Yuri Cristelli de Sousa Silva

1Center for Technology in Molecular Medicine (CTMM), Federal University of Minas Gerais (UFMG), Belo Horizonte, Minas Gerais, Brazil

1, Ana Caroline Nogueira Souza

Ana Caroline Nogueira Souza

2Post‐graduate Program in Molecular Medicine, Federal University of Minas Gerais (UFMG), Belo Horizonte, Minas Gerais, Brazil

2, Eduardo de Souza Nicolau

Eduardo de Souza Nicolau

2Post‐graduate Program in Molecular Medicine, Federal University of Minas Gerais (UFMG), Belo Horizonte, Minas Gerais, Brazil

2, Débora Marques Miranda

Débora Marques Miranda

3Department of Pediatrics, Federal University of Minas Gerais (UFMG), Belo Horizonte, Minas Gerais, Brazil

3, Marco Aurélio Romano‐Silva

Marco Aurélio Romano‐Silva

4Department of Psychiatry, Federal University of Minas Gerais (UFMG), Belo Horizonte, Minas Gerais, Brazil

4,✉

• Author information
• Article notes
• Copyright and License information

1Center for Technology in Molecular Medicine (CTMM), Federal University of Minas Gerais (UFMG), Belo Horizonte, Minas Gerais, Brazil

2Post‐graduate Program in Molecular Medicine, Federal University of Minas Gerais (UFMG), Belo Horizonte, Minas Gerais, Brazil

3Department of Pediatrics, Federal University of Minas Gerais (UFMG), Belo Horizonte, Minas Gerais, Brazil

4Department of Psychiatry, Federal University of Minas Gerais (UFMG), Belo Horizonte, Minas Gerais, Brazil

*

Correspondence:

Marco Aurélio Romano‐Silva (romano-silva@ufmg.br)

✉

Corresponding author.

Revised 2025 Oct 25; Received 2025 Jun 13; Accepted 2025 Oct 27; Issue date 2025 Nov.

© 2025 The Author(s). Journal of Neurochemistry published by John Wiley & Sons Ltd on behalf of International Society for Neurochemistry.

This is an open access article under the terms of the http://creativecommons.org/licenses/by/4.0/ License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited.

PMC Copyright notice

PMCID: PMC12617393  PMID: 41236438

ABSTRACT

The implementation of touchscreen platforms in co‐clinical trials for rodents (i.e., mice and rats) and humans to assess cognitive functions presents an opportunity to overcome barriers present in conventional clinical trials. To better visualize the progress made in this area, this review proposes a systematic synthesis of the comparability of touchscreen cognitive assessment studies applied to both humans and rodents in a co‐clinical framework. To accomplish this objective the Ovid, PubMed, Scopus and ScienceDirect databases were searched, in English, and without publication date limit and registered on the International Prospective Register of Systematic Review (PROSPERO) under the number CRD420250650537. The screening resulted in 5 cross‐sectional studies and 1 randomized controlled trial (RCT) included, which were assessed for methodological quality and risk of bias using the Joanna Briggs Institute (JBI) critical appraisal tools. The data acquired in this review reinforce the potential of touchscreen platforms for cognitive assessment across human and rodent models. Behavioral flexibility and visuospatial cognition excelled in terms of comparability. The scarcity of studies and methodological diversity represent significant gaps in the field. Regardless, the available data highlight important opportunities for advancing translational research in cognition with a co‐clinical approach.

Keywords: co‐clinical, cognition, human, rodent, systematic review, touchscreen platform

---

The implementation of touchscreen platforms in co‐clinical trials for rodents (i.e., mice and rats) and humans to assess cognitive functions presents an opportunity to overcome barriers present in conventional clinical trials. A systematic review was conducted to assess the comparability of the use of touchscreen cognitive assessment in a co‐clinical framework addressing humans and rodents. The data acquired reinforce the potential of touchscreen platforms for cognitive assessment. Behavioral flexibility and visuospatial cognition excelled in terms of comparability. The available data highlight important opportunities for advancing translational research in cognition with a co‐clinical approach.

---

Abbreviations

5C‐CPT

5‐choice continuous performance test

APL

acute promyelocytic leukemia

BF

Bayes Factor

CANTAB

Cambridge Neuropsychological Test Automated Battery

CIRP

Co‐Clinical Imaging Research Resource Program

D‐amp

dextroamphetamine

DLG2

discs large homolog 2

EMOTICOM

emotional and social function battery

HD

Huntington’s disease

IED

intra‐extra dimensional set‐shifting

JBI

Joanna Briggs Institute

KO

knock‐out

MoCA

Montreal Cognitive Assessment

NCI

National Cancer Institute

OCD

obsessive‐compulsive disorder

PAL

paired‐associate learning

PECO

Population, Exposure, Comparator, and Outcomes

PR

progressive ratio

PRISMA

Preferred Reporting Items for Systematic Reviews and Meta‐Analyses

PROSPERO

Prospective Register of Systematic Review

RCT

randomized controlled trial

RVP

rapid visual information processing

SWM

spatial working memory

TUNL

trial‐unique nonmatching to location

WT

wildtype

1. Introduction

Co‐clinical trials have emerged as an innovative approach by enabling real‐time data integration between experiments conducted in humans and animal models, promoting greater accuracy and relevance in the results obtained (Nardella et al. 2011). This study modality aims to solve the delays present in the usual preclinical‐clinical model and was initially used in the search for a cure for acute promyelocytic leukemia (APL), with successful results, as a “preclinical‐clinical” approach (Nardella et al. 2011; Balasubramanian et al. 2021). Although it is a field of current interest, a notable peak in publications and citations in this field occurred between 2013 and 2015, coinciding with increased support from the National Cancer Institute (NCI) for precision medicine, and its Co‐Clinical Imaging Research Resource Program (CIRP) (Dumont et al. 2021; Zhang 2023). However, there is a lack of systematic literature reviews on co‐clinical studies to support decision‐making and support researchers in the neuroscience field.

The distribution of resources for conducting co‐clinical research is not homogeneous around the world; therefore rodents (i.e., mice and rats) are generally preferred when conducting animal research (Carr 2025). Although nonhuman primates share remarkable similarities to humans due to shared evolutionary history, their use in research has been dwindling because of maintenance costs, difficult access to specialized labor, several ethical reasons, and the “3 Rs” of research (Replacement, Reduction, and Refinement), aimed at replacing animals with alternative models whenever possible, reducing the number of animals used, and refining procedures in order to minimize suffering (Harding 2017; Lopresti‐Goodman and Villatoro‐Sorto 2022). Rodents, however, can be cheaper to maintain and can be housed in smaller facilities; they are extensively used due to their physiological and genetic similarities to humans, as well as their shorter life cycle, which allows for faster observation of biological effects and more reliable anticipation of expected clinical results in humans (Carr 2025; Soufizadeh et al. 2024).

Conventional studies have traditionally been conducted in a fragmented manner, with isolated preclinical and clinical trials. This leads to methodological discrepancies between experiments, incomparable results, and challenges to develop new therapies (Salmon et al. 2024). This translational disconnection is a major barrier in the development of therapies targeting neuropsychiatric conditions (Salmon et al. 2024). Despite numerous studies aimed at the development of new treatments, many drugs have failed in clinical trials (Reuben et al. 2024; Granzotto et al. 2024). In neuroscience, one of the reasons for this high failure rate is the difficulty in having animal models that faithfully represent human psychiatric symptoms, both from a biological perspective and in the behavioral tasks used for assessment (Lee et al. 2021).

There is a discrepancy in the application of cognitive tests in rodents and those used in humans. While rodents are often evaluated using mazes, humans are assessed using broader tools such as the computerized tasks from the Cambridge Neuropsychological Test Automated Battery (CANTAB) or paper‐and‐pencil tests, which cover multiple domains of cognition (Lee et al. 2021). Recent advances, such as automated touchscreen platforms free from experimenter interference, have reduced stress and variability, allowing for more accurate and reproducible behavioral analyses (Huang et al. 2023; Horner et al. 2013). Touchscreen‐based tests emerge as a tool that offers greater precision and standardization, allowing humans and rodents to respond to visual stimuli in a similar manner (Bussey et al. 2012; Horner et al. 2013). While human participants touch the screen with their finger, rodents interact with the stimulus by touching with their snout (Palmer et al. 2021). The adaptation of CANTAB‐inspired paradigms to rodents increased the translational validity of the tests, enabling direct comparisons with human assessments (Palmer et al. 2021). Paired‐Associates Learning (PAL), for example, assesses associative memory and, in aged rats, revealed changes in connectivity between the hippocampus, prefrontal cortex, and the retrosplenial region, suggesting preservation of neural plasticity (Gaynor et al. 2023). Tasks such as PAL and Trial‐Unique Nonmatching to Location (TUNL), when applied intensively, also improved cellular plasticity and cognitive performance in Alzheimer’s models, resulting in increased synaptogenesis and improved cognitive phenotype (Shepherd et al. 2021). These tests have also been useful for identifying deficits after central nervous system injuries, such as trauma or stroke, with high accuracy in assessing memory, cognitive flexibility, and learning (Cotter et al. 2023). The integration of touchscreen technology with features such as electroencephalograms and optogenetics has enabled the simultaneous recording of behavior and neural activity, with high temporal resolution and minimal manual interference. This has enabled detailed correlations between neurophysiological variables and behavioral measures such as reaction time, number of errors, and exploration patterns (Kangas et al. 2021; Piantadosi et al. 2025).

The benefits of using touchscreen technology in rodents have facilitated the investigation of specific cognitive processes by neuroscientists from different specialties, using high standardization, minimal experimenter interference, and strong translational potential (Horner et al. 2013). The translational value stems from the similarity between the tasks employed in both rodents and humans, which confers a degree of face validity. Face validity refers to how much a task appears to measure the same psychological construct across species, based on similarities in procedures and structure. This differs from construct validity, which assesses whether the task truly measures the intended cognitive process, and from predictive validity, which examines whether task performance can forecast real‐world or clinically relevant outcomes. While face validity alone is not sufficient to guarantee these other forms of validity, it increases the likelihood of achieving them when tasks are designed to mimic human paradigms closely (Bussey et al. 2012). These advantages have contributed to the growing popularity and relevance of touchscreen‐based assessments in modern neuroscience (Dumont et al. 2021).

Because they generate highly consistent data over time, touchscreen platforms are especially suitable for integrating advanced experimental methodologies, benefiting both basic neuroscience research and drug discovery efforts. Furthermore, their potential for integration with state‐of‐the‐art neural recording technologies makes these platforms particularly effective in multiplexed studies (Carr 2025). However, for co‐clinical trials with touchscreen devices to be successful, standardization is necessary. To this end, the community must synchronize its efforts, and inspiration can be drawn from the success of this approach in the field of oncology. Seeking to integrate preclinical and clinical research, we observed the need to understand the possibilities of using touchscreens, given their sensitivity to detect cognitive deficits in animal models, especially in rodents. Considering the reality of population aging and the increase in neurodegenerative and neuropsychiatric diseases, the demand for viable techniques for conducting simultaneous co‐clinical studies in this area became evident. These studies provide, in addition to diagnostic accuracy, practical applicability and ethical obviousness in the study of cognition. In addition, co‐clinical studies that applied touchscreens can show ways to maximize the use of touchscreens in translational protocols and strengthen the scientific basis that supports their application in clinical practice. For this reason, we conducted a systematic review with the objective of identifying primary studies that reported synchronous co‐clinical investigations using touchscreens for cognitive assessment in humans and rodents.

2. Method

A systematic review was conducted adhering to the Preferred Reporting Items for Systematic Reviews and Meta‐Analyses (PRISMA) (Page et al. 2021) and registered in the International Prospective Register of Systematic Reviews (PROSPERO), under the number CRD420250650537, based on the guiding question “how comparable are the results for cognitive tests using touchscreen platforms in neuropsychiatric disorders in a co‐clinical approach for humans and rodents?”. The population, exposure, comparator, and outcomes (PECO) strategy was adopted, and the population consisted of humans and mice or rats, exposed to the touchscreen platform, compared to their respective control groups. The outcome expected was the comparability in cognitive function evaluation.

From the guiding question the following descriptors were used: (touchscreen operant system) AND (rat) OR (mice) AND (translational) OR (translational science) OR (translational research) AND (humans) AND (cognitive test) AND (cognition) in all fields on the Ovid, PubMed and Scopus databases, and in titles, keywords and abstract on ScienceDirect, in English and without publication date limit. Citation search in gray literature was not performed.

The inclusion criteria were experimental articles that investigate cognitive processes in mice/rats and humans using touchscreen technology, written in English. The exclusion criteria set were nonexperimental studies and preprints, studies in other languages aside from English, studies that investigated only rats/mice or only humans, studies that used a platform other than touchscreen for the mice/rats, studies with any other animal that are not mice/rats, and studies with nonhuman primates. The decision to exclude nonhuman primates was based on careful methodological and practical consideration. Rodent models are well established and widely validated in the scientific literature, serving as a robust reference for preclinical studies. Evidence consistently shows that rats and mice produce results that are highly compatible with human findings across multiple fields, including pharmacology, behavioral neuroscience, and studies of neuropsychiatric disorders. Given this strong translational potential, we focused on assessing whether such compatibility also holds in the emerging context of touchscreen‐based cognitive testing.

2.1. Screening Procedure

The screening was conducted through peer review using the Rayyan (Ouzzani et al. 2016) platform to organize, manage, and screen the articles. Initially, duplicate articles were eliminated as described in Figure 1 (PRISMA fluxogram). Conflicts between the reviewers regarding eligibility were discussed and resolved by consensus with a third member. Afterwards, the selected articles were read in full, and only those that effectively met the established criteria were included.

FIGURE 1.

PRISMA fluxogram.

2.2. Data Extraction

Data were extracted using a standardized spreadsheet that included the following data: first author, year of publication, country, study type, article objective, sample characteristics, such as age, sex, and species, type of touchscreen intervention including the name of the test battery and the specific task, and outcome measures. The data extraction process was conducted in pairs.

2.3. Quality Assessment

The methodological quality and risk of bias of the included studies were assessed using the “Critical Appraisal Tools” developed by the Joanna Briggs Institute (JBI) for cross‐sectional (Aromataris et al. 2024) and randomized controlled trials (RCTs) (Barker et al. 2023). Rather than relying on a numerical scoring system, this tool provides conceptual guidance to support the reviewer’s judgment. Based on this qualitative assessment, studies were categorized as having good, moderate, or poor methodological quality, depending on the presence and severity of design or implementation flaws.

Using a critical and qualitative approach, cross‐sectional studies were rated as high quality when all appraisal criteria were met; as moderate quality when items related to confounding factors (Q5 and Q6) were inadequately addressed; and as low quality when other key methodological elements (Q1, Q4, and Q8) were missing (see Table 3). Regarding RCTs, studies were considered high quality when all criteria were fulfilled; moderate quality when concerns were raised about randomization (Q1, Q6, and Q13) or allocation procedures (Q2); and low quality when other critical methodological domains (Q3, Q4, Q5, Q7, Q9, Q10, and Q11) were not adequately addressed (see Table 4). The complete table with the questions included, is available as Data S1.

TABLE 3.

Quality assessment for cross‐sectional studies.

Author (year)	Q1	Q2	Q3	Q4	Q5	Q6	Q7	Q8	Quality
Benzina et al. ( 2021)	★	★	★	★	★	★	★	★	High
Chow et al. ( 2020)	★	★	★	★	★	★	★	★	High
Heath et al. ( 2019)	★	★	★	★	★	☆	★	★	Moderate
Nithianantharajah et al. ( 2013)	★	★	★	★	☆	☆	★	★	Moderate
Nithianantharajah et al. ( 2015)	★	★	★	★	☆	☆	★	★	Moderate

Note: ★, yes; ☆, no.

TABLE 4.

Quality assessment for randomized controlled trials.

Author (year)	Q1	Q2	Q3	Q4	Q5	Q6	Q7	Q8	Q9	Q10	Q11	Q12	Q13	Quality
MacQueen et al. ( 2018)	★	★	★	★	★	☆	★	★	★	★	★	★	☆	Moderate

Note: ★, yes; ☆, no.

3. Results

See Figure 1.

3.1. Global Distribution

The six selected articles were published between 2012 and 2021, comprising four cross‐sectional studies and one randomized clinical trial. Collectively, they include 372 human participants and an estimated sample of 293–353 mice. The global distribution is presented in Table 1.

TABLE 1.

Distribution by country.

Country	Number of studies	Human subjects	Mice sample
Australia	1	140	45
France	1	80	52
UK	3	148	96–106 a
USA	1	4	100–150 b
Total	6	372	~293–353

Note: ~ Estimated value based on previous data from the table.

a

Estimated value based on the methodology described by Nithianantharajah et al. ( 2015).

b

Estimated value based on the methodology described by Nithianantharajah et al. ( 2013).

3.2. Study Characteristics

Most studies (66.7%) used transgenic or knockout mice, them being Sapap3 knock‐out mutant mice ( Sapap3 KO) (Benzina et al. 2021), R6/1 (Heath et al. 2019) and Dlg1+/−, Dlg2−/−, Dlg3−/y, and Dlg4−/− (Nithianantharajah et al. 2013, 2015). Only two studies did not use wildtype (WT) mice as controls, also being the only studies to use C57BL/6 mice (MacQueen et al. 2018), and one of them had the mice modeled for photothrombotic stroke (Chow et al. 2020). Studies predominantly (83.4%) evaluated a condition, either genetic (Nithianantharajah et al. 2013, 2015) or health‐related (Benzina et al. 2021; Heath et al. 2019; Chow et al. 2020). A single study had its main goal on the outcome after pharmacological intervention (MacQueen et al. 2018).

3.3. Participants and Animal Subjects’ Characteristics

Participants’ ages ranged from 20 to 81 years (weighted mean 60.8 pooled standard deviation 13.5) and animal subjects’ ages ranged from 1.4 to 7 months (weighted mean 3.3 pooled standard deviation 0.2). In the pooled participant sample, there was a nearly equal participation of males (45.9%), though one study did not provide complete gender data for all human samples (Benzina et al. 2021). In the pooled animal sample, only male subjects were included. Data from the included studies are detailed in Table 2.

TABLE 2.

Characteristics of the studies included in the review.

Author (year), country, study type	Objective	Sample, intervention	Findings
Benzina et al. ( 2021), France, cross‐sectional	To investigate behavioral flexibility in compulsive behaviors in obsessive‐compulsive disorder (OCD) patients and Sapap3 KO mice	Humans, N: 80 N: 40 control/OCD Mean age (SD): 40.28 (13.59) control, 40.15 (13.22) OCD Gender: 15 males control/OCD Intervention: Reversal learning task Mice, N: 52 N: 26 WT/ Sapap3 KO Age in days (SD): 200.12 (12.23) WT, 199.35 (12.29) Sapap3 KO Gender: all males Intervention: Reversal learning task	OCD or Sapap3 KO gene did not have an impact on performance (Bayes Factor (BF)Inclusion < 1 for group factor). There was not a difference in the number of trials to reach the reversal criterion (BF10 = 0.64, d = 0.27 [0.04 0.77]), nor mice (BF10 = 0.7, d = 0.33 [−0.1 0.92]). There was no group difference in the number of reversal errors (humans: BF10 = 1.5, d = 0.35 [0.07 0.88]; mice: BF10 = 0.44, d = 0.25 [−0.2 0.83]). OCD “checkers” displayed more behavioral inflexibility (BF+0 = 4.66, d = 0.74 [0.08 1.25]). Impaired Sapap3 KO presented more behavioral inflexibility than unimpaired Sapap3 KO mice and WT controls (BF10 > 100, d = 1.37 [0.54 2.17]). There were differences of spontaneous strategy change probability between the three subgroups (humans: BF10 = 1.19, η 2 = 0.08; mice: BF10 > 100, η 2 = 0.41).
Chow et al. ( 2020), Australia, cross‐sectional	To evaluate poststroke cognitive function in human chronic stroke survivors and photothrombotic stroke mice model	Humans, N: 140 N: 70 stroke/nonstroke Mean age (SD): 61.9 (13.8) stroke, 64.6 (10.0) nonstroke Gender: 38 males stroke, 24 males nonstroke Intervention: CANTAB, PAL task Mice, N: 45 N: 20 sham, 24 stroke Age: 7–8 weeks Gender: all males Intervention: PAL task	Stroke survivors made fewer correct responses for the first time, in comparison to participants within the control group ( p = 0.002). Mice modeled for stroke had worse performance in comparison to the sham group ( p = 0.032). There was a significant inverse association between stroke and the first attempt memory score was present in humans (crude β (95% CI), −2.26 (−3.62 to −0.899), p = 0.001), as for mice there was a significant effect of stroke ( F(1, 18) = 5.65; p = 0.029) and time ( F(6, 108) = 15.1; p < 0.0001) on the mean correct rate. Stroke survivors had a higher number of attempts to successfully complete an attempt ( p = 0.015). Mice modeled for stroke needed longer times to complete a session ( p < 0.0001) and completed a reduced number of tasks ( p < 0.0001).
Heath et al. ( 2019), England, cross‐sectional	In this present study they aimed to demonstrate the validity of using touchscreen‐delivered progressive ratio tasks to mirror apathy assessment in Huntington’s disease (HD) patients and a representative mouse mode	Humans, N: 43 N: 23 HD, 20 control Mean age (SD): 53.6 (14.6) HD, 52.2 (20.2) control Gender: 13 HD males, 10 control males Intervention: Emotional and social function battery (EMOTICOM) Mice, N: 52 N: 23 R6/1, 29 WT Age: 6 weeks Gender: all males Intervention: Progressive ratio	Motivation was higher for controls in both species (humans: p < 0.001, d = 1.6; mice: p < 0.001; d = 1.09), supported by post reinforcement pause (humans: p < 0.001, d = 1.4; mice: p < 0.001; d = −1.53). There was a significant reduction in R6/1 animals relative to WT littermates, corresponding to a lower intrinsic motivational baseline ( p < 0.001; d = 1.96). For humans, the highest number of touchscreen responses in the patient group was highly correlated to functional decline in everyday activities and lower scores indicated worse impairment ( r = 0.45, p < 0.05).
MacQueen et al. ( 2018), England, RCT	To investigate how D‐amp would improve 5C‐CPT performance in a similar manner for both healthy adult mice and human participants, demonstrating pharmacological validity for the task	Humans, N: 71 Age: 18–35 years Gender: 47.9% male Intervention: 5‐choice continuous performance test (5C‐CPT) Mice, N: 32 Age: ~8 weeks at onset of testing Gender: all males Intervention: 5C‐CPT	For humans, D‐amp significantly improved performance at both the 10 and 20 mg doses, relative to placebo ( d = 0.821 and 0.758; p < 0.05), such a result was also seen at the 0.3 mg/kg in mice relative to saline ( d = 1.138). The effect was driven by increased hit rate in both species (humans: p = 0.001, d = 0.938 and 0.903; mice: p = 0.017, d = 1.197) and concurrent reduced percent omissions in humans ( p = 0.001; d = 0.897 and 0.807) at both doses, and for mice at the smaller dose, although it did not reach statistical significance ( p = 0.055). The number of correct responses was also improved in the two species, by both doses of D‐amp for humans ( p < 0.001; d = 1.115 and 1.076) and at the 0.3 mg/kg dose of D‐amp relative to saline for mice ( p = 0.002; d = 1.676). In mice improved accuracy only happened at the 0.3 mg/kg dose, while the 1.0 mg/kg reduced accuracy ( d = 0.845). Hit reaction‐time variability was, however, reduced by both doses ( p < 0.001; d = 1.001 and 1.338) for humans, and for mice it had no significant effect ( p = 0.859).
Nithianantharajah et al. ( 2013), USA, cross‐sectional	To examine the genetic basis of the vertebrate cognitive repertoire through paralogous genes and compare homologous cognitive processes in mice and humans	Humans, N: 4 Age: 24–67 years old Gender: 1 male Intervention: Spatial working memory (SWM), intra‐extra dimensional set‐shifting (IED), PAL, Rapid Visual Information Processing (RVP) Mice, N: ~10–15 for each cohort N: Dlg1+/−, Dlg2−/−, Dlg3−/y, Dlg4−/− and WT Age: undisclosed Gender: all males Intervention: Pavlovian conditioned approach, visual discrimination and reversal learning task, object‐location paired‐associates learning task, extinction, five‐choice serial reaction time task (5‐CSRTT)	Mice with Dlg2−/− and humans with mutations in discs large homolog 2 (DLG2) made significantly more errors than healthy control subjects from the general population in tests of visual discrimination acquisition and cognitive flexibility ( p < 0.005) and visuo‐spatial learning and memory ( p < 0.005). Humans with mutations in DLG2 also showed decreased accuracy compared to controls in a test for sustained attention ( p < 0.005), similar to the impaired response accuracy seen in Dlg2−/− mice.
Nithianantharajah et al. ( 2015), England, cross‐sectional	To assess mice and humans carrying disease‐related genetic mutations with an identical touchscreen based cognitive test	Humans, N: 34 N: 4 DLG2, 30 controls Age: 23–67 years old Gender: all females DLG2, 14 male controls Intervention: Adapted rodent touchscreen object‐location paired associates learning test Mice, N: ~10–15 per group N: ~10–15 Dlg2−/−/WT Age: undisclosed Gender: all males Intervention: Rodent touchscreen object‐location paired associates learning test	Dlg2−/− mice presented a striking impairment in object‐location paired associates learning, performing consistently on an average of 50% chance level on all three blocks of trials ( p < 0.01). Human controls displayed progressive acquisition of object‐location paired associates across training trials ( p < 0.001), akin to WT mice, without gender or IQ making a difference on performance ( p = 0.74). Individuals with DLG2 CNV deletions tested on the same test failed to show this progressive acquisition ( p = 0.549), still performing approximately at chance level.

3.4. Quality Assessment

See Tables 3 and 4.

3.5. Findings

In Benzina et al. ( 2021) results were similar between humans and mice, when comparing performance profiles in a reversal event in subjects with OCD and controls. Similar performance profiles could be observed after a reversal event between compulsive subjects and their controls in the two species, meaning OCD or the knockout of the Sapap3 gene did not have an impact on performance (BFInclusion < 1 for group factor). The number of trials needed to reach reversal criterion did not differ between compulsive and control groups, neither for human subjects (BF10 = 0.64, d = 0.27 [0.04 0.77]), nor for mice (BF10 = 0.7, d = 0.33 [−0.1 0.92]). Similarly, no significant group differences were found in the number of reversal errors, neither in humans (BF10 = 1.5, d = 0.35 [0.07 0.88]), nor in mice (BF10 = 0.44, d = 0.25 [−0.2 0.83]). In OCD patients, correlation analysis showed that disease severity and task performance were not related. Likewise, there was no correlation in mice between grooming level and the main behavioral parameters. A post hoc analysis revealed that OCD “checkers” needed more trials than both OCD “noncheckers” (BF+0 = 4.66, d = 0.74 [0.08 1.25]) and healthy controls (BF+0 = 9.32, d = 0.67 [0.14 1.17]), therefore displaying more behavioral inflexibility. Similarly, “impaired” Sapap3 KO presented more behavioral inflexibility than “unimpaired” Sapap3 KO mice and WT controls (BF10 > 100, d = 1.37 [0.54 2.17]). There was a positive correlation between the severity of “checking” symptoms and the probability of spontaneous strategy change, that is, changing its response despite positive feedback. These results suggested that OCD “checkers” had a high response lability as identified through an elevated spontaneous strategy change probability. The subgroup analysis in both species supported this result, with differences in spontaneous strategy change probability observed between the three subgroups either in humans (BF10 = 1.19, η 2 = 0.08) or in mice (BF10 > 100, η 2 = 0.41).

Similarly, in the study by Chow et al. ( 2020), individuals affected by stroke demonstrated notably fewer correct responses when attempting to recall the appropriate patterns for the first time, in comparison to participants within the control group ( p = 0.002). Likewise, mice that experienced a stroke displayed poorer performance, achieving a markedly lower accuracy rate in comparison to the sham group ( p = 0.032). For humans, a significant inverse association between stroke and the first attempt memory score was present (crude β (95% CI), −2.26 (−3.62 to −0.899), p = 0.001), as for mice there was a significant effect of stroke ( F(1, 18) = 5.65; p = 0.029) and time ( F(6, 108) = 15.1; p < 0.0001) on the mean correct rate, meaning that stroke considerably affects the capacity to retain and manipulate visuospatial information, supported by the increased number of attempts to successfully complete each level by stroke survivors ( p = 0.015), similarly observed by mice’s longer times to complete a session ( p < 0.0001), still completing a reduced number of tasks ( p < 0.0001).

When investigating the use of touchscreen‐delivered progressive ratio tasks as an assessment tool for apathy in Huntington’s disease, Heath et al. ( 2019) found that, assessed by breakpoint value, motivation was higher for controls in both species (humans: U = 394.0, p < 0.001, d = 1.6; mice: t(48.162) = 4.0879; p < 0.001; d = 1.09). Post reinforcement pause corroborated this finding (humans: U = 56.0, p < 0.001, d = 1.4; mice: t(29.606) = −5.0861; p < 0.001; d = −1.53). Once again, the comparison of the estimated peak progressive ratio response rate indicated a significant reduction in R6/1 animals relative to WT littermates, corresponding to a lower intrinsic motivational baseline ( p < 0.001; d = 1.96). For humans, the highest number of touchscreen responses in the patient group was highly correlated to functional decline in everyday activities and lower scores indicated worse impairment ( r = 0.45, p < 0.05).

MacQueen et al. ( 2018) investigated the effects of D‐amp on the 5C‐CPT in humans and mice. For humans, D‐amp significantly improved performance at both 10 and 20 mg doses, relative to placebo ( d = 0.821 and 0.758; p < 0.05); such a result was also seen at 0.3 mg/kg in mice relative to saline ( d = 1.138). The effect was driven by an increased hit rate in both species (humans: F(2, 136) = 7.628, p = 0.001, d = 0.938 and 0.903; mice: F(3, 28) = 4.015, p = 0.017, d = 1.197) and concurrent reduced percent omissions in humans ( F(2, 68) = 7.350, p = 0.001; d = 0.897 and 0.807) at both doses, and for mice at the smaller dose, although it did not reach statistical significance ( F(3, 28) = 2.852, p = 0.055). The number of correct responses was also improved in the two species, by both doses of D‐amp for humans ( F(2, 136) = 12.037, p < 0.001; d = 1.115 and 1.076) and at the 0.3 mg/kg dose of D‐amp relative to saline for mice (F(3,28) = 6.012, p = 0.002, d = 1.676). Although D‐amp improved accuracy for humans at both doses, in mice it only happened at the 0.3 mg/kg dose, while the 1.0 mg/kg red

(Truncated: full text at source URL.)