Running Head: PROSPECTIVE MEMORY IN VR AND REAL LIFE 1 *** This manuscript draft is under review. Do not cite without authors’ permission! *** Immersive virtual reality gameplay captures age differences in naturalistic prospective memory Joseph Saito & Nathan S. Rose* University of Notre Dame *Address Correspondence to: Nathan S. Rose 390 Corbett Family Hall Notre Dame, IN 46556 [email protected] PROSPECTIVE MEMORY IN VR AND REAL LIFE 2 Abstract Age differences in prospective memory (PM)—memory for delayed intentions—have shown paradoxical patterns between laboratory and naturalistic settings. Virtual reality (VR) has been used to try and enhance the ecological validity of PM assessments, but methodological differences and limited validation have undermined interpretation of previous findings. We compared age differences between VR- and naturalistic-based measures of PM performance for younger (18-30 years) and older (56-83 years) adults (N = 111) to explore the role of task context and familiarity. Participants completed PM tasks embedded in the Job Simulator VR videogame and a Breakfast task that involved setting a table and simulating breakfast food preparation. We also included two real-world measures in which participants tried to remember to exchange personal belongings with the experimenter (Belongings task) and return phone calls at specific times outside the lab (Call-back task). We found comparable age deficits in Job Simulator and the Breakfast task. However, the age-PM paradox persisted in the Belongings and Call-back tasks. Hierarchical regression modeling was conducted to determine the roles of working memory, vigilance, and personality traits in each. Regression analyses revealed that significant variance in lab-based PM performance was accounted for by individual differences in working memory and agreeableness in older adults, while vigilance and neuroticism in young adults accounted for variance in naturalistic PM performance. This study suggests that immersive VR gameplay provides ecologically valid PM assessment and advances a theoretical account of the age-PM paradox with a systematic, task-based analysis of age and individual differences in PM. Keywords: prospective memory, aging, virtual reality, naturalistic, vigilance, working memory, personality PROSPECTIVE MEMORY IN VR AND REAL LIFE 3 Prospective memory (PM) refers to the collection of cognitive processes that allow individuals to remember to perform intended actions at the appropriate moment in the future. Daily life is critically dependent on PM (e.g. remembering to buy milk at the market, taking medication at the proper time) (Einstein & McDaniel, 1990). As such, the preservation of PM in later stages of life is critical for maintaining an independent lifestyle (Crovitz & Daniel, 1984; Terry, 1988; Einstein & McDaniel, 1990; Ellis, 1996; Kliegel & Martin, 2003; Hering et al., 2018; Woods et al., 2012). A substantial body of research has been conducted to try and understand the origins of PM differences between younger and older adults (Brandimonte, Einstein, & McDaniel, 2014). The multi-process framework of PM was developed by McDaniel & Einstein (2000) to describe how strategic monitoring and spontaneous retrieval processes support PM differently as a function of task features, such as the type of retrieval cue, as well as the differences between groups of individuals, such as younger and older adults. The Multi-process Framework of Prospective Memory PM tasks with event-based and time-based PM cues have been shown to place different levels of demand on individuals’ cognitive abilities (Will et al., 2009; Rose et al., 2010; Aberle et al., 2010; Rose et al., 2015). Event-based tasks are triggered by an event in one’s environment, and have typically displayed better performance compared to time-based tasks—which consist of intentions cued by the passage of a set period of time. This benefit for event-based tasks is suggested to result from decreased demand on the controlled attentional and working memory processes required for active monitoring and retrieval (Einstein et al., 1995; Park et al., 1997; Brandimonte, Ferrante, Feresin, & Delbello, 2001; d’Ydewalle, Bouckaert, & Brunfaut, 2001; Cona, Bisiacchi, & Moscovitch, 2013; Rose et al., 2015). Older adults have been shown to display significant age-related impairments on time-based task performance relative to younger PROSPECTIVE MEMORY IN VR AND REAL LIFE 4 adults (d’Yedwalle et al., 2000; Einstein et al., 1995; Park et al., 1997; Rendell & Craik, 2000). While individuals display age decrements in event-based tasks as well, the size of group differences can be mitigated or eliminated by adjusting the salience of the retrieval cue (McGann, Defeyter, Ellis, & Reid, 2005; Rose et al., 2010). A cue is more salient when it is centrally-embedded in the ongoing task, making it easier to detect via spontaneous retrieval processes (McDaniel & Einstein, 2000). This is in line with Craik, Klix, & Hagendorf’s (1986) hypothesis that older adults show greater deficits on tasks requiring self-initiated monitoring and retrieval. However, age differences in PM performance are not consistent across studies. Older adults are sometimes found to perform comparably or even outperform younger adults—a pattern referred to as the age-prospective memory paradox (Rendell & Thomson, 1999; Rendell & Craik, 2000). The Age-Prospective Memory Paradox While age-related declines in PM are typically observed in conventional lab-based paradigms, older adults have been shown to perform comparably, and sometimes outperform younger counterparts when the task is to be performed in naturalistic settings (Aberle et al., 2010; Rendell & Thomson, 1999; Rendell & Craik, 2000; Kliegel, McDaniel & Einstein, 2008). One hypothesis is that this reversal of age-related deficits is due to differences in the ecological validity of naturalistic tasks (e.g. having subjects simulate adhering to a medication regimen in real life over the course of several days) and conventional lab-based measures of PM (e.g. pressing a computer key when a specific word is presented during a computerized lexical decision task). It may be the case that the latter does not capture PM processes as they operate in everyday life. This concern about ecological validity is common with regards to psychological assessment. Parsons’ (2015) review describes many traditional psychological assessments as PROSPECTIVE MEMORY IN VR AND REAL LIFE 5 “construct-driven” rather than “functionally-driven”, meaning that attempts to isolate psychological constructs may counterproductively limit prediction of real-world behavior. As it pertains to age differences in PM research, few studies have systematically compared differences across multiple different lab-based and naturalistic task settings. While lab-based measures enable greater experimental control than naturalistic measures (Einstein and McDaniel, 1990; Einstein, Holland, McDaniel, & Guynn, 1992), their abstract, unfamiliar attributes may disproportionately impact older adults compared to younger adults (Altgassan, Kliegel, Brandimonte, & Filippello, 2010). Unfortunately, attempts to use naturalistic task designs that are more familiar to older adults make it difficult to account for differences in lifestyle demands and motivation between age groups (Henry et al., 2004; Rendell & Thomson, 1999, Aberle et al., 2010). As a result of these limitations, a comprehensive account of the age- PM paradox is still lacking. In an attempt to bridge the gap in ecological validity between naturalistic and lab-based measures of PM, research using virtual reality (VR) has become more prominent in recent years. Virtual Reality and Prospective Memory Virtual reality, a technology developed and marketed largely as an entertainment product, has been utilized to induce naturalistic perceptual experiences without costs in experimental control. Not surprisingly, the rapid evolution of VR in recent years has created inconsistency in the way individuals define “virtual reality” when describing their methods. These methodological differences are important because they have implications about the extent to which VR is able to simulate real-world interaction. In the case of the extant PM literature, visual stimuli are predominantly presented via either standard computer monitors (Debarnot et al., 2015; Gonneaud et al., 2012; Sakai et al., PROSPECTIVE MEMORY IN VR AND REAL LIFE 6 2018) or head-mounted displays (HMD) (Parsons & Barnett, 2017; Brown et al., 2016; Banville et al., 2010). While each of these methods can make use of three-dimensional environments, they are critically different in the way they embody them. HMDs allow for visual displays that adjust in real-time with physical head movements. The duality of synchronized visual input and motor function has been shown to recreate real-world patterns of neural activation to a greater extent than visual input alone (Taube, Valerio, & Yoder, 2013). One study used virtual reality with mice and found that only 25% of localized place cell activation was achievable with visual input alone, while the remaining 75% activation required additional proprioceptive and vestibular information (Chen, King, Burgess, & O’Keefe, 2012). Another drawback cited in Rizzo, Buckwalter, & Neumann’s (1997) review of VR’s use in cognitive rehabilitation argued that “cyber sickness”, a form of motion sickness resulting from perceptual “lag” between visual and motor modalities was a significant barrier in the expansion of VR. Fortunately,