Abstract

INTRODUCTION

Animal studies have consistently shown that exploration of a novel environment can promote long-term potentiation (LTP) in the hippocampus and specifically in the dentate gyrus, thereby improving memory encoding (Li et al., 2003; Straube et al., 2003a; Davis et al., 2004; Sajikumar and Frey, 2004; Sierra-Mercado et al., 2008). The modulatory effect of novelty on learning has been suggested to rely on the functional connections between the hippocampus and the substantia nigra/ventral tegmental area (SN/VTA), components of the mesolimbic dopaminergic system, by regulating the entry of new information into long-term memory (Lisman and Grace, 2005). When novelty is detected, a novelty signal from the hippocampus activates dopaminergic neurons in the SN/VTA. This results in dopamine being released in the hippocampus, where dopamine increases plasticity by enhancing LTP. Interestingly, increased plasticity not only occurs during exploration of the novel environment, but lasts up to 10 min afterwards (Li et al., 2003; Straube et al., 2003a; Kentros et al., 2004).

Direct evidence for a link between exploration of novel environments and increased plasticity comes from animal studies, but also in humans evidence for such a link has been found. Active as compared to passive exploration of a virtual environment (VE) has been shown to improve recall of allocentric spatial information (Voss et al., 2011; Plancher et al., 2012, 2013). Memory improvements have been observed for familiar items when these were presented in the context of novel scene stimuli (Bunzeck and Düzel, 2006). Similarly, passive exposure to novel scenes has been shown to improve recall of words presented in a task after exposure (Fenker et al., 2008). In this latter study either novel or previously familiarized scenes were presented for which indoor/outdoor judgments had to be made. After exposure to novel compared to familiar scenes recall and recollection on an unrelated word learning task were superior, whereas no such improvements were observed for familiarity-based recognition (Fenker et al., 2008). Also memory improvements have been found for items that stand out in the to-be-learned list, a benefit that may be novelty-related. For example, in the Von Restorff or distinctiveness effect memory formation is better for words presented in distinctive novel fonts than for words presented in a standard font (Von Restorff, 1933; Bruce and Gaines, 1976; Schmidt, 1985; Hunt, 1995; Dunlosky et al., 2000; Geraci and Rajaram, 2004; Geraci and Manzano, 2010; Rangel-Gomez and Meeter, 2013). Furthermore, the anticipation of a novel stimulus has been shown to enhance memory encoding for unexpected novel events (Wittmann et al., 2007). However, no studies so far have looked at whether active exploration of a novel environment promotes memory relative to familiar environments, or whether the effects of exploration extend to a subsequent learning task in humans.

In the present study, participants performed a word learning task either after exploring a previously familiarized or a novel VE from a first-person perspective in a within-subjects design (see Figure ​Figure11). In line with the findings above, we expected learning to be improved after exploring a novel VE, relative to a familiarized one. As was found by Fenker et al. (2008) and given that exploration of spatial novelty has been strongly associated with the hippocampus (Lisman and Grace, 2005; Jeewajee et al., 2008; Bast et al., 2009; Kaplan et al., 2012), we expected a learning benefit for free recall, which is thought to depend on the hippocampus (Yonelinas et al., 2002), and not for recognition, believed to rely mainly on the perirhinal cortex (Aggleton and Brown, 1999; Wan et al., 1999; Duzel et al., 2001; Fernandez et al., 2002; Yonelinas et al., 2002).

Open in a separate window

FIGURE 1

Experimental design and apparatus. (A) In session 1 participants were familiarized either with Tuscany or the Residential virtual environment (VE). In session 2 they either explored the familiarized VE (Familiar First) or the novel VE (Novel First). In the final session 3, participants in the Novel First condition explored the familiar VE, and participants in the Familiar First condition explored the novel VE. (B) Participants explored a VE and filled in the Igroup Presence Questionnaire (IPQ) in all sessions. In sessions 2 and 3 they also performed the word learning task. (C) Shows an example binocular display of the Tuscany VE, and the Oculus Rift head-mounted display.

MATERIALS AND METHODS

RESULTS

Figure ​Figure22 shows average performance on the word learning task after exploring both a novel and familiar environment. Recall was better after participants explored the novel versus the familiar VE, F(1,24) = 4.01, p = 0.046, = 0.16. Whether participants explored the novel or familiar VE in the first session did not affect recall (F < 1), nor did it matter which VE was familiarized, F(1,24) = 1.45, p = 0.240, = 0.06. There was a trend for an interaction between Novelty and which VE was familiarized, F(1,24) = 3.19, p = 0.087, = 0.12, with effects of novelty being stronger when Tuscany than when Residential was familiarized. Other factors did not interact with Novelty (p > 0.554).

Open in a separate window

FIGURE 2

Memory performance on the word learning task. Recall memory, accuracy as defined by d’ on the recognition test, and response bias on the recognition test after exploring a novel and familiar environment. Error bars reflect standard errors.

Whether the novel or familiar VE was explored or which type of VE was explored first did not affect sensitivity during recognition (F < 1). There was a trend for higher performance for the participants that were familiarized with Tuscany rather than Residential, F(1,24) = 4.10, p = 0.054, = 0.15. No interactions with Novelty were found (F < 1). In addition, no differences in response bias were found after exploring a novel or familiar VE (F < 1). Novelty did not interact with which VE was familiarized (F < 1), nor with whether participants first explored the novel or familiar VE, F(1,24) = 1.47, p = 0.237, = 0.06.

Presence ratings (IPQ) were higher after exploration of the novel (mean = 7.32; SD = 13.41) than of the familiar VE (mean = 3.79; SD = 11.44), t(27) = 2.11, p < 0.045. In addition, recall memory was better for participants who gave high compared to low presence ratings, F(1,26) = 4.96, p = 0.035, = 0.16. Presence did not interact with Novelty, F(1,20) = 2.29, p = 0.146, = 0.10. No effects of presence were found for recognition (neither hits nor false alarms; F < 1). To investigate whether novelty’s effects on recall were the result of higher presence ratings in the novel VE, linear regression analysis was performed for the novelty recall benefit with the presence difference score as the predictor. The presence difference score was not related to the novelty recall benefit, β = -0.002, t(27) = -0.865, p = 0.395, and only marginally lowered the effect of novelty on recall, suggesting that presence and novelty independently affected recall. Figure ​Figure33 shows the recall scores after exploration of a novel or familiar environment and the corresponding IPQ presence scores.

Open in a separate window

FIGURE 3

Recall for low and high presence. The proportion of correctly remembered words after exploration of the novel and familiar environments relative to the respective IPQ scores (Presence).

DISCUSSION

The effect of exploring of a novel versus a familiar environment was investigated in humans using virtual reality environments and performance on a subsequent word learning task. Recall was enhanced for words learned after exploration of a novel compared to a familiar environment. No such effect was found for recognition. Recall is believed to rely strongly on the hippocampus, while recognition depends more on the perirhinal cortex and thalamus (Aggleton and Brown, 1999; Wan et al., 1999; Fernandez et al., 2002). Therefore, these findings are in line with the Lisman and Grace’s (2005) model that proposes that a novelty signal from the hippocampus may promote subsequent memory encoding by dopamine release from the SN/VTA projecting back to the hippocampus. Familiarity-based recognition does not strongly rely on the hippocampus, which means that the Lisman and Grace (2005) model does not predict an effect of novelty on recognition. Indeed no effects were found on recognition. In contrast, Bunzeck and Düzel (2006) reported enhanced recognition memory for familiar items when these were presented in a block in which also novel stimuli were presented, relative to when they were presented among even more familiar items. Their finding is difficult to compare to the present study’s results, since their measure of recognition was corrected hit rate, subtracting false alarms from hits. This procedure does not fully disentangle changes in accuracy from criterion shifts, which leaves open the possibility that their results reflect a shift in the response criterion rather than improvements in memory.

Exposure to novelty has been reliably shown to improve memory in non-human animals (Li et al., 2003; Straube et al., 2003a; Davis et al., 2004; Sajikumar and Frey, 2004; Sierra-Mercado et al., 2008), as has exploration (Li et al., 2003; Straube et al., 2003a). In humans, memory is better for distinctive items in a list than for non-distinctive items. This so-called Von Restorff effect (Von Restorff, 1933) has sometimes been taken to reflect an effect of novelty on learning (Kishiyama et al., 2004, 2009), though some authors have argued that other mechanisms are responsible for this effect (Bruce and Gaines, 1976; Schmidt, 1985; Hunt, 1995; Dunlosky et al., 2000; Geraci and Rajaram, 2004; Rangel-Gomez et al., 2012; Rangel-Gomez and Meeter, 2013). Our study, and that of Fenker et al. (2008) are the only ones that report an effect in humans similar to those reported in non-human animals: temporally extended memory enhancements after exposure to novelty. In our study, exploration of novel environments improved learning in an unrelated task that was on average 10 min after the exploration phase. Temporally extended novelty-induced memory enhancements may be related to the mesolimbic dopamine system. Increased dopaminergic drive in the hippocampus may promote encoding of new information especially after novelty detection and has been associated with a oscillatory theta state in the hippocampus (Lisman and Otmakhova, 2001). Indeed, activation of dopaminergic D1/D5 receptors during exploration of a novel environment has been found to lower the threshold for LTP and learning in the hippocampus in rats (Li et al., 2003). Also in humans, novelty triggers dopamine release in the SN/VTA, which results in better memory performance (Schott et al., 2004; Wittmann et al., 2007; Apitz and Bunzeck, 2013; Eckart and Bunzeck, 2013). Dopamine is known to result in increased plasticity in the hippocampus (Chen et al., 1999; Lisman et al., 2011), and the effects of novelty on dopamine release are known to affect plasticity 10 min later (Li et al., 2003; Straube et al., 2003a; Kentros et al., 2004).

Long-lasting effects may also partly be mediated by motivation. Dopamine release may result in improved motivation by eliciting an “exploration bonus” (Kakade and Dayan, 2002; Lisman and Grace, 2005; Bunzeck and Düzel, 2006; Knutson and Cooper, 2006; Bunzeck et al., 2007, 2012; Krebs et al., 2009, 2011; Duzel et al., 2010; Guitart-Masip et al., 2010), which may also have contributed to enhanced recall memory. This is supported by the fact that recall was higher for participants who reported a stronger sense of presence, and that participants gave higher presence ratings for the novel than for the familiar environment. Presence has been linked to attention allocated to the virtual rather than the real world (Darken et al., 1999), and may thus correlate with how motivated participants were during the exploration. This could possibly have contributed to the novelty-induced memory enhancements. However, since differences in presence in the novel and familiar VE were not related to the novelty recall benefit it cannot explain the novelty effect. Presence thus promotes memory via a relatively independent mechanism.

Alternatively, the observed learning benefits may be caused by increased arousal as mediated by noradrenergic activity. Exploration of novel environments results in arousal by increasing noradrenergic activity (Moser et al., 1994; Sara et al., 1994; Vankov et al., 1995; Kitchigina et al., 1997; Straube et al., 2003b), enhancing LTP and memory formation (Klukowski and Harley, 1994; Cahill and McGaugh, 1998). Also the cholinergic system has been linked to novelty’s effect on learning (Hasselmo, 1999; Ranganath and Rainer, 2003; Meeter et al., 2004; Barry et al., 2012); acetylcholine has been shown to be present in cortex in higher levels during exploration of a novel rather than a familiar environment (Giovannini et al., 2001). Noradrenergic, cholinergic and dopaminergic neuromodulatory systems could each underlie the novelty-induced memory enhancements, alone or in combination; direct physiological evidence linking these systems to memory performance is still lacking.

In conclusion, our findings suggest that active exploration of a novel environment can promote memory encoding also in humans, and that these effects linger for some time after exploration. In addition, individual differences and fluctuations in the amount of attention allocated to the environment during exploration also affect learning in the later memory task.

Conflict of Interest Statement

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Acknowledgments

Footnotes

REFERENCES