Individual variability in visual discrimination and reversal learning performance in common marmosets
Introduction
Common marmosets (Callithrix jacchus) have recently been used as nonhuman primate models for various human diseases in neuroscience research (Mansfield, 2003, ‘t Hart et al., 2012, Kishi et al., 2014). Efficient and stable methods to assess cognitive functions are essential in such models. Visual discrimination and reversal learning have often been used in marmoset research (Cotterman et al., 1956, Ridley et al., 1981, Ridley et al., 1986, Ridley et al., 1993, Roberts et al., 1988, Roberts et al., 1992, Dias et al., 1996, Fine et al., 1997, Harder et al., 1998, Smith et al., 1999, Clarke et al., 2004, Clarke et al., 2005, Clarke et al., 2007, Clarke et al., 2008, Clarke et al., 2011, Pryce et al., 2004, Man et al., 2009, Rygula et al., 2010), and they are applicable to a wide range of investigations of different processes in reinforcement learning, such as attention, value estimation, and memory formation (Izquierdo and Jentsch, 2012, Gilmour et al., 2013, Klanker et al., 2013). In rodent studies, visual discrimination and reversal learning have been widely used, and the established protocols enable new researchers to easily initiate their experiments (e.g., Horner et al., 2013). In marmoset studies, the use of visual discrimination and reversal learning has been limited to a small number of laboratories, likely because detailed information about experimental procedures and characteristics of learning behavior in marmosets have not been reported. To our knowledge, no previous studies have systematically examined individual variability in the learning performance of marmosets.
In our laboratory, over 40 marmosets have thus far been trained in visual discrimination and reversal learning. We attempted to quantify their performance. To this end, appropriate measures for the animals’ performance are necessary. The most commonly used measure is the total number of trials or errors until the animal achieves a certain criterion. However, we did not adopt this measure in the present study, as a criterion has typically been determined arbitrarily and such a measure has therefore differed across studies. Instead, we adopted more objective and informative measures based on a learning curve, which represents the probability of a correct response as a function of trial number.
Currently, there are two popular methods to assess performance from the learning curve for each subject (Gallistel et al., 2004, Smith et al., 2004), both of which have advantages over conventional methods (e.g., the moving average method). One is a model-based method proposed by Smith et al. (2004) in which a state-space smoothing algorithm is used and the estimated learning curve is gradual. The other is a purely descriptive method proposed by Gallistel et al. (2004) in which change points are determined on a cumulative response curve by a recursive algorithm and the estimated learning curve is given as a staircase function. As the latter method is sensitive to an early onset and abrupt changes in learning curves, the authors found that different onsets and abrupt rises can be detected in individual learning curves, even in the gradual group-average learning curve. In the present study, we adopted the method of Gallistel et al. (2004) and used two representative values—onset trial and dynamic interval—as measures of individual learning performance.
To determine factors that affect performance, we examined the effects of age, sex, and experimental circumstance (in a colony or an isolator rack) on performance. These factors would be considered during experiment planning and subject selection. In addition, we compared the performance of subjects from different families to examine whether a family factor might affect learning behavior.
Section snippets
Subjects
Forty-two naïve adolescent and young adult common marmosets (C. jacchus; 24 males and 18 females, aged 1 year, 7 months to 4 years, 3 months) were used in this study (Table 1). Subjects were born and reared at the Primate Research Institute, Kyoto University. Behavioral experiments were conducted in an individual cage placed in a colony or an isolator rack system (Natsume Seisakusho Co. Ltd., Japan) between 13:00 and 16:00 on weekdays. Subjects were fed 30 g of New World monkey pellets once
The ratio of successfully learned marmosets
Of 42 marmosets, 37 (88%) learned to touch a visual stimulus on the screen in the pre-training (Table 1). Two of the five dropouts (nos. 38 and 39) would not touch the stimulus at the initial stage of the pre-training for more than 10 days. The remaining three dropouts (nos. 40–42) actually touched the stimulus but could not improve their touching methods in the 2-week pre-training. Thus, we trained 37 marmosets in visual discrimination learning, in which all marmosets succeeded. We then
Discussion
In this study, we quantified the performance of more than 40 adolescent and young adult marmosets in visual discrimination and reversal learning. We successfully elucidated some characteristics of marmoset learning ability.
As visual discrimination and reversal learning are often used to assess prefrontal functions, including perseveration and impulsive behavior, the learning statistics presented here are useful for many researchers. Based on these data, all marmosets that learned to touch the
Conclusions
Almost all marmosets could complete visual discrimination and reversal learning. Only one marmoset failed to complete reversal learning. They showed very high performance, with over 95% correct responses in 87% of measurements, and over 90% in 98% of measurements. Such high performance can make the experimental system for marmosets sensitive to the effects of experimental manipulations. With two measures independent of arbitrary criteria, we successfully quantified their learning processes. As
Acknowledgements
This work was partially carried out under the SRPBS from MEXT to KN. This work was partially supported by Cooperative Research Projects for marmoset research to KN.
References (36)
- et al.
Structure learning in action
Behav. Brain Res.
(2010) Metalearning and neuromodulation
Neural Netw.
(2002)- et al.
Learning impairments following injection of a selective cholinergic immunotoxin, ME20.4 IgG-saporin, into the basal nucleus of Meynert in monkeys
Neuroscience
(1997) - et al.
Measuring the construct of executive control in schizophrenia: defining and validating translational animal paradigms for discovery research
Neurosci. Biobehav. Rev.
(2013) - et al.
The role of the central cholinergic projections in cognition: implications of the effects of scopolamine on discrimination learning by monkeys
Brain Res. Bull.
(1998) - et al.
Deprivation of parenting disrupts development of homeostatic and reward systems in marmoset monkey offspring
Biol. Psychiatry
(2004) - et al.
Stimulus-bound perseveration after frontal ablations in marmosets
Neuroscience
(1993) - et al.
Learning impairment following lesion of the basal nucleus of Meynert in the marmoset: modification by cholinergic drugs
Brain Res.
(1986) - et al.
A specific form of cognitive rigidity following excitotoxic lesions of the basal forebrain in marmosets
Neuroscience
(1992) - et al.
Meta-learning in reinforcement learning
Neural Netw.
(2003)
The dopamine D3/D2 receptor agonist 7-OH-DPAT induces cognitive impairment in the marmoset
Pharmacol. Biochem. Behav.
Development of a compact and general-purpose experimental apparatus with a touch-sensitive screen for use in evaluating cognitive functions in common marmosets
J. Neurosci. Methods
The marmoset monkey: a multi-purpose preclinical and translational model of human biology and disease
Drug Discov. Today
Increase in reaction time for solving problems during learning-set formation
Behav. Brain Res.
Cognitive inflexibility after prefrontal serotonin depletion
Science
Dopamine, but not serotonin, regulates reversal learning in the marmoset caudate nucleus
J. Neurosci.
Lesions of the medial striatum in monkeys produce perseverative impairments during reversal learning similar to those produced by lesions of the orbitofrontal cortex
J. Neurosci.
Prefrontal serotonin depletion affects reversal learning but not attentional set shifting
J. Neurosci.
Cited by (21)
The marmoset as a model for investigating the neural basis of social cognition in health and disease
2022, Neuroscience and Biobehavioral ReviewsCitation Excerpt :A possible contributory factor to females’ preference to engage in habitual behaviour could be differences in the OFC/striatal dopaminergic system, which is an area that can be influenced by circulating oestrogen levels (Lacreuse et al., 2014; LaClair and Lacreuse, 2016; Shams et al., 2016; Brzezinski-Sinai and Brzezinski, 2020; Veselic et al., 2020). Interestingly, both Takemoto et al. (2015) and Ash et al. (2020) reported that there were no sex differences in reversal learning in adolescent marmosets. The discrepancy between the previous studies and the Takemoto and Ash studies can be attributed to the latter studies using younger marmosets.
Sex differences in cognitive aging: a 4-year longitudinal study in marmosets
2022, Neurobiology of AgingCitation Excerpt :Reversal learning is a test of executive function that assesses cognitive flexibility, or the ability to adapt to changing reward-stimulus contingencies. Reversal learning has been well studied in NHPs including marmosets (Kangas et al., 2016; LaClair and Lacreuse, 2016; Lacreuse et al., 2014a; Munger et al., 2017; Pryce et al., 2004; Ridley et al., 1981; Sadoun et al., 2019; Strasser and Burkart, 2012; Takemoto et al., 2015) and involves a network of brain regions including the orbitofrontal cortex, caudate and striatum (Clarke et al., 2005; Clarke et al., 2011; Collins et al., 2000; Jackson et al., 2018; Roberts et al., 1990). In year 1, a subset of monkeys were also tested on intradimensional/extradimensional set shifting (n = 17) (Laclair et al., 2019), and progressive ratio (n = 16) (Carlotto, unpublished results) but these tests were not administered in subsequent years and are not reported here.
Bayesian inference for variability discrimination on partial sameness
2019, NeurocomputingKey periods of cognitive decline in a nonhuman primate model of cognitive aging, the common marmoset (Callithrix jacchus)
2019, Neurobiology of AgingCitation Excerpt :OT is thus a learning parameter that reflects the number of trials needed to exceed the chance level. The AT was the trial from which an asymptotic level of performance was reached (Takemoto et al., 2015). The dynamic interval (DI) was considered as the number of trials between OT and AT, corresponding to a measure of a transition from a chance level with unconditioned state to a conditioned learning.
Development of stereotaxic recording system for awake marmosets (Callithrix jacchus)
2018, Neuroscience ResearchCitation Excerpt :The present stereotaxic recording system is suitable for neuronal recording during task performance of limb and oculomotor movements, since marmosets can move their hands freely and their fixed-heads allow us to easily track their eye movements. Marmosets can learn not only simple motor tasks (Tia et al., 2017) but also various types of tasks to assess cognitive functions, such as visual discrimination (Maclean et al., 2001; Takemoto et al., 2015), reversal learning (Clarke et al., 2005; Rygula et al., 2010; Takemoto et al., 2015) and decision-making (Tokuno and Tanaka, 2011). Thus, neural mechanisms underlying motor control and cognitive functions can be investigated in marmosets by using the present stereotaxic recording system.
Visual discrimination and reversal learning in aged common marmosets (Callithrix jacchus)
2017, Neuroscience Research
- 1
These authors contributed equally to this work.