PROJECT DESCRIPTION

​

  Scrub jays encode information about what is where and when, monkeys can report when they lack information, ravens plan for a future event, and great apes can even anticipate the behavior of others that possess false beliefs (Shettleworth, 2010). Nonhuman animals represent the world they perceive in sophisticated ways, but it is unclear whether they can also represent imaginary situations, an ability that endows humans with remarkable cognitive prowess. Could some of these remarkable problem-solving demonstrations be grounded on the representation of imagined scenarios? Our goal is to investigate whether nonhuman primates use representations of the virtual world 1) in a similar way to the representations of the real world, 2) to enhance their learning and planning abilities in the real world (and vice versa) and even 3) to mentally travel between virtual and real worlds. Next we justify our choice of research domain (spatial knowledge), multiple species (monkeys, apes and humans), and innovative methods (Virtual Reality) in addressing the challenge of Diverse Intelligences.

  Research domain:  Research on spatial knowledge has played a key role in uncovering animal intelligences. Spatial and temporal variability in resources and intra- and inter-specific competition likely drive the diversity of intelligences in the foraging behavior of multiple species. Marmosets and chimpanzees travel between distantly located but synchronously fruiting trees that are ephemeral and rare in their forested landscape using memory and planning of relatively direct travel routes (Garber, 1989; Janmaat et. al. 2013; 2014). The cognitive requirements for foraging are tightly allied with processing of social information (Cunningham & Janson 2007; Rosati, 2017) so that the ecological and social intelligence hypotheses of cognitive evolution may be complementary (Overington et al., 2008). By focusing on spatial knowledge, we maximize our ability to capture diverse intelligences because we 1) ask questions about ecological and social intelligence, 2) assess intelligence defined as computation of novel information derived from prior knowledge and experience - a perceptual restructuring of knowledge (Dolins & Mitchell, 2010), and the various related operations that develop including memory, inference, planning, equivalency, social coordination and communication, and 3) can study multiple species. 

  Multiple species: Capturing the richness of diverse intelligences requires the study of multiple species (Call et al., 2017).  Our selection of chimpanzees, bonobos, humans and rhesus macaques is strategic in providing a relatively broad taxonomic coverage, and crucially, some insight into the effects of language training, computer-related experience and cultural upbringing on those abilities. Elucidating the origin and diversity of intelligence requires careful inter-specific comparisons of multiple species and intra-specific comparisons that shed light on ontogenetic influences.  If humans solve spatial problems more flexibly and accurately than nonhuman primates, is this due to possessing better representations due to language usage? Does being raised within a language-rich environment, permit certain cognitive capacities to emerge in captive nonhuman primate species (by affording opportunities not accessible in the wild) (e.g., Levinson, 2003)?  Some of the apes included in this study were raised in a language-rich environment and can communicate with humans using lexigrams on a computer touchscreen (link to Kanzi at the lexigram keyboard; Brakke & Savage-Rumbaugh,1995; Hillix & Rumbaugh, 2004; Lyn et al., 2011; Beran & Heimbauer, 2015). Other apes and monkeys included in this study are not lexigram-trained and so provide useful comparison groups. Rhesus monkeys have demonstrated considerable cognitive skills on computerized tasks including mazes (Washburn & Astur, 2003; Beran et al., 2015), video-tasks (Rumbaugh et al., 1989), matching-to-sample (Hopkins et al., 2002), numerosity (Brannon & Terrace, 2000), and psychophysical switching tasks (Avdagic et al., 2014), although they have not yet been tested using virtual reality (VR) as proposed to do. Their data will provide a broader comparison of spatial skills among apes and monkeys, and a direct comparison between traditional computerized displays and our current VR platform. Our human participants will include adults and children from three distinct cultural and linguistic groups: suburban United States, and in the Republic of Congo, the Bantu (farmers) and Mbendjele (hunter-gatherers). The Bantu and Mbendjele live in a shared village, with the Mbendjele dispersing into the forest to obtain food and other resources, while the Bantu farm nearby land. The Bantu speak Lingala, and their children attend school, read and write. By contrast, the Mbendjele speak their own language (Mbendjele) practice oral traditions, are not literate, do not traditionally use writing, and have no experience with modern technology (e.g., cars, planes, television, phones, computers). Focusing on multiple species maximizes our ability to capture diverse intelligences by 1) providing sufficient taxonomic coverage and, 2) allowing us to investigate the effects of language training, computer-related experience and cultural upbringing on ecological and social cognition.

  Virtual Reality: We borrow the methods used in Artificial Intelligence and design a virtual world that subjects can navigate to obtain food items using touchscreen and joystick technology. As they move through the virtual environments, the view will shift accordingly, simulating real world perception. We use the term "virtual reality" (VR) rather than "3D video game" for sake of consistency with recent publications adopting this naming convention (Chouinard-Thuly et al., 2016; Dolins, et al., 2017). Although the subject is viewing three-dimensional objects on a flat two-dimensional screen rather than through a binocular headset, the essential feature of interactive first person immersion is maintained. Unlike classical computerized tasks, which typically present simple static stimuli (e.g., pairs of photographs or geometric figures) and require long training regimes (Shettleworth, 2010), we confront subjects with a much more complex stimulus array, both in terms of the number of elements present on the screen and the way they move and change positions. It may seem paradoxical to want to simplify the task by using more complex stimuli, but we expect that when those stimuli represent real-life scenarios (e.g., a landscape that moves as the subject navigates) this approach may be key to uncovering capabilities unobserved to date. There are several cases where making the task harder by increasing the number of steps, helped subjects succeed (Girndt et al., 2008; Schubiger et al. 2016). One great advantage of VR is that it provides experimental control and extreme flexibility, the opportunity to present simulated built or naturalistic environments, and to present the same experimental paradigms to multiple primate species and individuals with widely different experiences and backgrounds (Dolins et al., 2014; 2017).  Since we cannot control previous experience in wild nonhuman primates, we do not know their specific knowledge base. With VR, we can control what they can know, what information they use to implement planning routes, and their navigational progress at choice points while still having the opportunity to create similar environments and challenges that primates face in their natural habitats. In preliminary test trials chimpanzees have been shown to readily adapt to this task.

  Research aims: We focus on navigation and goal-directed foraging in a virtual world, investigating whether individuals remember food locations, visit the sites at an appropriate time, and outsmart potential competitors in order to forage efficiently in the virtual world. With increasing complexity aimed at emulating the challenges posed by space, time and social agents that primates routinely face in the real world our tests will assess multiple aspects of intelligence including memory, learning, integration, inference, anticipation, planning, and deception. We also focus on equivalency of virtual and real worlds: understanding symbolic representation - investigating the question of equivalency more directly by asking to what extent objects and locations in the virtual world are treated like symbolic representations of their real counterparts. Our series of experiments will test various aspects of this relation including analogical reasoning, iconicity and symbol use.

 Our ultimate goal is to gather knowledge that inspires a general interest and sense of wonderment about the scientific process, and what new ideas it can uncover, stretching the imagination to unsuspected boundaries.