Assessing cognitive flexibility in humans and rhesus macaques with visual motion and neutral distractors

Abstract

IntroductionCognitive flexibility is the ability of an individual to make behavioral adjustments in response to internal and/or external changes. While it has been reported in a wide variety of species, established paradigms to assess cognitive flexibility vary between humans and non-human animals, making systematic comparisons difficult to interpret.MethodsWe developed a computer-based paradigm to assess cognitive flexibility in humans and non-human primates. Our paradigm (1) uses a classical reversal learning structure in combination with a set-shifting approach (4 stimuli and 3 rules) to assess flexibility at various levels; (2) it employs the use of motion as one of three possible contextual rules; (3) it comprises elements that allow a foraging-like and random interaction, i.e., instances where the animals operate the task without following a strategy, to potentially minimize frustration in favor of a more positive engagement.Results and DiscussionWe show that motion can be used as a feature dimension (in addition to commonly used shape and color) to assess cognitive flexibility. Due to the way motion is processed in the primate brain, we argue that this dimension is an ideal candidate in situations where a non-binary rule set is needed and where participants might not be able to fully grasp other visual information of the stimulus (e.g., quantity in Wisconsin Card Sorting Test). All participants in our experiment flexibly shifted to and from motion-based rules as well as color- and shape-based rules, but did so with different proficiencies. Overall, we believe that with such approach it is possible to better characterize the evolution of cognitive flexibility in primates, as well as to develop more efficient tools to diagnose and treat various executive function deficits.

A Touchscreen-Based, Multiple-Choice Approach to Cognitive Enrichment of Captive Rhesus Macaques (Macaca mulatta)

“…On average animals conducted 418 trials per session (Q1 = 239, Q3 = 806; Figure 2a), with no evidence of a significant decrease between the first and the last session (assessed through an animal-wise partial correlation that accounted for variations in session duration, Table 1). Within sessions, however, we observed a decrease in trial number (two-sided t-test; p-value = 0.0115, t = −2.58, N = 92), indicating that animals engaged more in the first half of the session compared to the second half (Figure 1c, see Methods 2.6.1 as well as previous publications [5,49,50]).…”

“…The eXperimental Behavioral Instrument XBI [4,5,28,49] is mounted directly on the animals' enclosure and can autonomously run cognitive experiments as well as various enrichment protocols (Figure 1). The device can operate stand-alone for several hours, requires little to no human supervision, needs minimal weekly maintenance, provides video monitoring and recording of each session, and is fully integrated into the DPZ's local area network.…”

“…To estimate the uniformity of engagement within each given session, across animals and sessions, we (1) normalized the time of each trial initiation to the time of each session end (Figure 2b) and (2) computed the median of each of the resulting distributions to (3) ultimately identify at which proportion of each session animals had performed half of the trials (Figure 2c, see also [5,49,50]). A distribution centered at 0.5 would indicate that animals performed as many trials before as after the session midpoint.…”