Journal of Experimental Child Psychology 130 (2015) 79–91
Contents lists available at ScienceDirect
Journal of Experimental Child Psychology journal homepage: www.elsevier.com/locate/jecp
What goes where? Eye tracking reveals spatial relational memory during infancy Jenny L. Richmond ⇑, Jenna L. Zhao, Mary A. Burns School of Psychology, University of New South Wales, Sydney, NSW 2052, Australia
a r t i c l e
i n f o
Article history: Received 25 November 2013 Revised 23 September 2014
Keywords: Spatial relational memory Binding Infancy Eye tracking Preferential looking Episodic memory
a b s t r a c t Episodic memory involves binding components of an event (who, what, when, and where) into a relational representation. The ability to encode information about the relative locations of objects (i.e., spatial relational memory) is a key component of episodic memory. Here we used eye tracking to test whether infants and toddlers learn about the spatial relations among objects. In Experiment 1, 9-, 18-, and 27-month olds were familiarized with an array of three objects. Following familiarization, they saw test arrays in which two of the objects had been replaced with novel ones (object switch condition) and arrays in which two of the objects had switched positions (location switch condition). Both 18- and 27-month olds looked significantly longer than would be predicted by chance at the objects that had switched spatial locations; however, 9-month olds did not. In Experiment 2, we showed that, given sufficient familiarization time, 9-month olds were also capable of detecting disruptions to the spatial relations among an array of objects. These results have important implications for our understanding of spatial relational memory development. Ó 2014 Elsevier Inc. All rights reserved.
Introduction Memories for events that we experience typically include details of who was there, what happened, when it happened, and where it happened. Compositionality is a key function of episodic memory ⇑ Corresponding author. Fax: +61 2 9385 3641. E-mail address: [email protected] (J.L. Richmond). http://dx.doi.org/10.1016/j.jecp.2014.09.013 0022-0965/Ó 2014 Elsevier Inc. All rights reserved.
80
J.L. Richmond et al. / Journal of Experimental Child Psychology 130 (2015) 79–91
(Henke, 2010); event memories are composed of individual components of an experience that are linked to form a coherent event representation. Episodic memories are composed of networks of relational representations (Cohen & Eichenbaum, 1993; Eichenbaum, Otto, & Cohen, 1992); relational networks allow event memories to be rapidly retrieved when a component cue is activated. Individual event networks are also connected to other events that share features, allowing us to make inferences about memories that are only indirectly related. Studies of lesioned animals (Eichenbaum, 2004), amnesic patients (Ryan, Althoff, Whitlow, & Cohen, 2000), and human brain activation (Konkel, Warren, Duff, & Tranel, 2008) have shown that the hippocampus is critically involved in episodic memory functions generally and memory for the relations among event components specifically. Relational memory refers to the binding of associated event elements, including individual items (who/what) as well as spatial (where) and temporal (when) information. Spatial relational memory allows us to remember ‘‘what’’ happens ‘‘where.’’ Remembering where you parked your car in the garage, or where your passport is in the filing cabinet, requires you to encode the position of the target object in relation to other objects in the environment. This kind of relational binding is essential for successful navigation and has been shown to depend on the hippocampus (Astur, Taylor, Mamelak, Philpott, & Sutherland, 2002; Banta Lavenex, Amaral, & Lavenex, 2006). Research looking at episodic memory development in toddlers has shown that children as young as 3 years are able to encode the relative location of objects in the environment and use this information to guide search behavior. For example, Hayne and Imuta (2011) had children hide three toys in specific locations in different rooms in their house (e.g., Big Bird is hidden in the bedroom behind the curtain). Following a 5-min delay, children were asked to verbally recall the identity and location of the toys along with the order in which they were hidden. Children were also asked to search for each object, and behavioral recall was coded. The results showed that although 3-year olds recalled less information than 4-year olds when asked about where the toys were hidden, spatial relational memory did not differ between the two groups. Both 3- and 4-year olds were able to accurately find the toys when asked to search for them. Spatial relational memory has also been studied in infants as young as 12 months using object search tasks. In these tasks, children watch the experimenter hide a toy and are encouraged to search for the toy after a delay (Bushnell, McKenzie, Lawrence, & Connell, 1995; Sluzenski, Newcombe, & Satlow, 2004). The accuracy of search behavior or the distance between the search position and hiding position (i.e., search error) is used as an index of memory. Children as young as 12 months can accurately search for a single object, particularly when its location is signaled by a distinctive cue; however, they have difficulty in using the relative location of two objects or cues to guide search behavior. For example, Bushnell et al. (1995) showed that 12-month olds were able to find a toy when it was hidden under a distinctive cushion, but not if it was hidden under a cushion that was in a particular location relative to a distinctive cushion. Similarly, Sluzenski et al. (2004) had children watch the experimenter repeatedly bury two toys that were always hidden in the same relative position to each other in a sandbox. Following a fixed training period, the experimenter hid the two toys out of sight of children, and then the location of one toy was revealed. Children were encouraged to search for the second toy using the relative location of the first one. Sluzenski and colleagues showed that 2- and 3-year olds were able to use spatial relations to accurately guide search behavior in this task; however, more than half of 18-month olds were unable to even complete the training. The ability to use the relative position of several cues to guide search behavior improves late in the second year of life (Ribordy, Jabès, Lavenex, & Lavenex, 2013; Sluzenski et al., 2004) and continues to improve during early childhood (Schutte, Spencer, & Schöner, 2003). Using a paradigm in which children searched an arena of overturned cups for hidden rewards, Ribordy et al. (2013) showed that it was not until 43 months of age that children reliably found rewards when searching in a complex array of 18 possible search locations. When only 4 possible search locations were used, 18- to 23-month olds still failed to reliably locate the cup containing the reward. Although these data, along with those of Sluzenski et al. (2004), suggest that spatial relational memory might not be functional during infancy, it is possible that infants learn about the spatial relations among objects before they are able to use this information flexibly to guide successful search behavior. Research using novelty preference tasks to study spatial relational memory in animals may be useful in informing our understanding of human development in this area.
J.L. Richmond et al. / Journal of Experimental Child Psychology 130 (2015) 79–91
81
Much evidence for the role of the hippocampus in spatial relational memory has been accumulated from animal studies using novelty preference tasks. These tasks capitalize on the animal’s natural tendency to explore novelty and allow similar procedures and dependent variables to be used across studies of rats and primates. Spontaneous preference tasks, such as the novel object recognition (NOR) task, are used in rat models. During the familiarization period, the rat is placed in an arena and allowed to explore two identical objects. The rat is removed from the arena, and two new objects are added; one object is identical to that seen during familiarization, and one is novel. The rat is placed in the arena again, and exploration is recorded. Memory is inferred if the rat exhibits a novelty preference, spending a greater percentage of time exploring the novel object than would be predicted by chance. Although object recognition is not affected by hippocampal lesions, when memory for the relative position of objects is tested, hippocampal rats exhibit impairments (Good, Barnes, Staal, McGregor, & Honey, 2007). For example, Good et al. (2007) tested rats with hippocampal lesions on the standard NOR task, in which familiar objects are replaced with novel ones at test, along with a version in which the relative location of objects was manipulated during the test. In this version, rats were familiarized with four objects and at test the location of two of the objects was switched. In the standard task, both hippocampal and control rats showed novelty preferences, exploring the novel objects more than the familiar ones. When tested on the spatial relational version, control rats spent a greater percentage of time than would be predicted by chance exploring the familiar objects that had switched locations; however, hippocampal rats did not explore the objects that had been switched for any longer than the objects that remained in the same configuration. These results suggest that the hippocampus is not critical for object recognition per se but is involved in integrating identity and location information into a relational network. Hippocampal animals are unable to encode the relative position of objects or recognize that the spatial relations in an array of objects have been changed (Good et al., 2007). Novelty preference tasks, like the visual paired comparison (VPC) task, have also been adapted to study spatial relational memory in primates. Bachevalier and Nemanic (2008) showed that monkeys with hippocampal lesions exhibit novelty preferences when tested on the VPC task with arrays that include novel objects, but not when tested on arrays in which the spatial relations among a set of objects have been manipulated. In this study, monkeys were familiarized with arrays of five objects and tested with arrays in which two or more of those objects had been replaced with novel ones (object identity condition) or arrays in which two or more of those objects had shifted relative position (object-in-place condition). The results showed that monkeys with hippocampal lesions exhibited novelty preferences when tested with object arrays containing novel items but exhibited null preferences when tested with arrays in which the identity of the objects remained the same but their relative position had changed. Blue, Kazama, and Bachevalier (2013) recently looked at the development of spatial relational memory in monkeys tested on the same task at 8 months, at 18 months, and as adults (5–6 years). In control animals, novelty preferences on this task increased with age; however, animals that had received neonatal hippocampal lesions exhibited null preferences at all ages. The ability to learn about spatial relations among an array of objects has a relatively protracted developmental trajectory and depends on the integrity of the hippocampus in monkeys (Bachevalier & Nemanic, 2008; Blue et al., 2013).
Experiment 1 Although novelty preference measures have been used extensively to study memory in human infants (Rose, Feldman, & Jankowski, 2004) and the extent to which infants form categories of spatial relations (Quinn, 2002), to date novelty preference measures have not been adapted to study spatial relational memory during infancy. Here we aimed to test whether 9- to 27-month olds learn about spatial relations by capitalizing on the spatial resolution of eye tracking. We chose this age range to include the youngest infants who have been shown to exhibit other kinds of relational memory (Richmond & Nelson, 2009) and to span the age range in which spatial relational memory has been demonstrated in search tasks (Sluzenski, Newcombe, & Kovacs, 2006).
82
J.L. Richmond et al. / Journal of Experimental Child Psychology 130 (2015) 79–91
There are several advantages to using eye tracking in this context. First, eye tracking allows us to test for spatial relational memory in infants who are motorically incapable of performing search tasks (i.e., 9-month olds). Second, there is much research to suggest that eye movements and gaze direction are critical in guiding search behavior (Filimon, 2010; Mushiake, Tanatsugu, & Tanji, 1997). Thus, by measuring eye movements directly, we are able to measure memory processes without the challenge of controlling for age-related differences in gross motor behavior. The procedure is adapted from the VPC task used by Bachevalier and Nemanic (2008), Blue et al. (2013); however, children were tested with single-object arrays rather than pairs of novel and familiar arrays. Children were familiarized with displays of three objects in a triangular spatial configuration (top left, top right, and bottom middle). On test trials, the familiar array was replaced with a novel array. In the object switch condition, two objects from the familiar array were replaced with new ones to create a novel stimulus. In the location switch condition, two objects from the familiar array switched locations to create a novel stimulus. If infants are sensitive to both the identity of objects and their spatial relations, we predicted that infants would look longer at the novel objects within the array on object switch trials and would look longer at the objects in the switched positions on location switch trials relative to the amount of looking that objects in these positions attracted during study. Alternatively, if children are unable to represent the spatial relations among objects until after 24 months of age, we predicted that 27month olds would look longer at the objects that have switched positions but that 9- and 18-month olds would not. Method Participants A total of 66 children were recruited from a database of families who had expressed an interest in participating in research; all were tested within 1 week of their birthday (9, 18, or 27 months). The 9-month olds (n = 22) were on average 276 days old (SD = 6), the 18-month olds (n = 24) were on average 547 days old (SD = 8), and the 27-month olds (n = 20) were on average 821 days old (SD = 4). There were 30 girls and 36 boys. Data from a total of 20 children was excluded due to fussiness (n = 8), insufficient looking on test trials (n = 4), calibration issues (n = 5), or technical issues (n = 3). The final sample included data from 16 9-month olds, 16 18-month olds, and 14 27-month olds. Parents were reimbursed for their travel costs, and children chose a small gift to take home as a ‘‘thank you’’ for participating. Apparatus and stimuli Fixation was recorded using a Tobii XL60 eye tracker. The eye tracker is composed of a 22-inch screen with a set of infrared cameras embedded in the perimeter. These cameras automatically capture a corneal reflection and, once calibrated, will track where the infant is looking on the screen. E-Prime software (Psychology Software) was used to present the stimuli and capture eye-tracking data. The stimuli were color photographs of toys (180 230 pixels, 11.7 12.5 cm on screen). In each block, children were presented with a group of three toys that were matched on the presence or absence of faces, color, and category (i.e., block toys, tower toys, or animal toys). There were a total of six triplet sets constructed. Procedure Following consent procedures, children were seated on their parent’s lap in front of the Tobii monitor. The position of the monitor was adjusted until the cameras detected the eyes, and a series of looming balls appeared at each of the corners of the screen and in the middle in a 5-point calibration sequence. The accuracy of calibration was checked and repeated if necessary. There were a total of 6 test blocks, each containing four 10-s study trials and two 10-s test trials (see Fig. 1). An attention-getter was presented before each trial to ensure that the infant was fixating the middle of the screen and that a good track was obtained before initiating each trial. During the study trials, three toys were presented on the screen in the top left, top right, and bottom middle positions. The configuration of the toys was consistent across the four study trials.
J.L. Richmond et al. / Journal of Experimental Child Psychology 130 (2015) 79–91
83
Fig. 1. Each block consisted of four 10-s study trials and two 10-s test trials. In the location switch condition (left panel), the position of two of the objects was switched at test. In the object switch condition (right panel), two objects were replaced with novel ones. Manipulated positions are marked for illustration only; the red ovals did not appear during stimulus presentation. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
In the first test trial, the three toys were presented again; however, two of the three toys were manipulated. On half of the trials, the position of two of the toys was switched (location switch condition). On the other half of the trials, two of the toys were replaced with novel ones (object switch condition). The positions that were manipulated were counterbalanced across trials (left–right, right–bottom, and left–bottom). On the second test trial, the original spatial configuration or identity of objects was restored. Blocks of trials continued until the infant was no longer interested in watching the pictures. There were a total of 6 blocks of stimuli, and 9-, 18-, and 27-month olds completed on average 5.63, 4.44, and 4.71 blocks, respectively. Results and discussion Consistent with previous work using eye tracking to measure relational memory (Richmond & Nelson, 2009; Richmond & Power, 2014), infants needed to spend at least 2500 ms1 fixating the stimuli during the test trial and contribute data on both location switch and object switch trials in order to be included in analyses. Areas of interest (top left, top right, and bottom middle) were defined a priori using inline code in the E-Prime script. The raw eye-tracking data were subjected to a fixation filter that determined the length of each fixation and the corresponding area of interest. The filter defined a fixation as a period of time in which the eye position does not move more than 50 pixels for at least 200 ms. Individ1 In our previous work using eye tracking to measure relational memory in infants, we required that infants spend 50% of the presentation time fixating the stimuli. Here we used 10,000-ms test trials; however, most infants spent only half of that time fixating the stimuli (M = 4870 ms, SD = 1824). We acknowledge that the trial length was likely too long, and rather than excluding more than half of the infants, we set the minimum test look criterion to be consistent with our previous work using a similar paradigm (Richmond & Nelson, 2009; Richmond & Power, 2014).
84
J.L. Richmond et al. / Journal of Experimental Child Psychology 130 (2015) 79–91
ual fixations were then analyzed to determine the percentage of time spent fixating the target positions during the study trials and the percentage of time spent fixating the manipulated positions during test. If attention is distributed equally among the objects in the left, right and bottom positions, we would expect infants to fixate objects in each position 33% of the time on average. During study, we predicted that infants would devote no more than 67% of the time to the ‘‘to be manipulated’’ objects. During test, if infants detected the spatial and/or nonspatial manipulations, we predicted that they would look disproportionately at the manipulated positions, fixating these objects significantly longer than would be predicted by chance (i.e., >67%). Analyses of variance (ANOVAs) were conducted separately for each condition (location switch and object switch) in order to test for differences in preferential looking as a function of age and trial phase. This approach is consistent with that applied in the original study using this method with primates (Bachevalier & Nemanic, 2008). Location switch condition Fig. 2 illustrates the mean percentages of time that 9-, 18-, and 27-month olds spent fixating the target objects (i.e., those in the switch positions) during the study and test phases of the location switch blocks. Here we analyzed the percentage of time spent fixating the target objects during the last familiarization trial relative to the first test trial using a repeated-measures ANOVA with age (9-month olds, 18-month olds, or 27-month olds) and trial phase (study or test) as factors. Data from the second test trial are not included in the analysis because the trial order was not counterbalanced and, as such, it is difficult to know to what extent infants’ looking behavior during this trial was influenced by the preceding test trial. This analysis revealed a main effect of trial, F(1, 43) = 7.11, p < .05, g2p = .14, but no main effect of group or Group Trial interaction (ps > .05). Averaged across age groups, infants spent a significantly greater percentage of time fixating the target objects that had switched locations during the test (M = 75, SD = 13) than during the final familiarization trial (M = 67, SD = 17). Although the ANOVA can tell us whether there is a difference in looking behavior as a function of age and/or trial phase, it does not speak to whether infants’ visual preferences differed from what would be expected by chance. One-sample t tests confirmed that during the final familiarization trial, infants in all age groups spent approximately 67% of the available looking time fixating the target objects. These preferences did not differ significantly from chance in 9-month olds, t(15) = 1.34, p > .05, 18-month olds, t(15) = 0.86, p > .05, or 27-month olds, t(13) = 0.08, p > .05. In contrast, during
Fig. 2. Location switch condition. Mean percentages of time that 9-, 18-, and 27-month olds spent looking at the target objects during the final study trial and first test trial are shown. In this condition, two of the three objects in the display switched locations between the study and test phases; the target objects were those that switched position. The dotted line represents chance performance of 67%. Looking that is equally distributed among the three positions would result in infants spending 67% of the time fixating the objects in the switched locations. Preferences significantly greater than chance are marked with an asterisk.
J.L. Richmond et al. / Journal of Experimental Child Psychology 130 (2015) 79–91
85
the first test trial, 18-month olds (M = 74), t(15) = 3.41, p < .05, and 27-month olds (M = 77), t(13) = 2.31, p < .05, looked significantly longer at the stimuli that had switched positions than would be predicted by chance. There was also a trend for 9-month olds to look preferentially at switched objects, (M = 75), t(15) = 2.12, p = .051. Object switch condition Fig. 3 shows the mean percentages of time spent fixating objects in the target positions during study and test trials. A repeated-measures ANOVA with age (9-month olds, 18-month olds, or 27month olds) and trial phase (study or test) as factors revealed no main effects or interactions (ps > .05). Much like the location switch trials, children spent on average 67% of the time fixating the target positions during the final familiarization trial. One-sample t tests showed that these preferences did not differ from chance (i.e., 67%) in 9-month olds, t(15) = 1.74, p > .05, 18-month olds, t(15) = 0.78, p > .05, or 27-month olds, t(13) = 0.55, p > .05. In the object switch condition, objects from two locations were replaced with novel objects during the first test trial. Surprisingly, there was no evidence of preferential looking at these novel objects during the test. The 9-month olds (M = 68), t(15) = 0.19, p > .05, the 18-month olds (M = 73), t(15) = 1.57, p > .05, and the 27-month olds (M = 73), t(13) = 1.21, p > .05, displayed preferences that did not differ significantly from chance, suggesting that they were not sensitive to changes in object identity. Familiarization time There is much evidence to show that the magnitude of infants’ novelty preferences is affected by the amount of study time given (Rose et al., 2004). In Experiment 1, infants were shown study configurations for a total of 40 s in both the location switch and object switch conditions; however, there may have been individual differences in the amount of familiarization time that was accumulated. For this reason, we wanted to rule out the possibility that differences in visual preference across age and condition could be accounted for by differences in initial study time. Table 1 contains the mean looking time accumulated during the familiarization period as a function of condition and age group. A repeated-measures ANOVA with condition (location switch or object switch) as a within-subjects variable and age group (9-month olds, 18-month olds, or 27-month olds) as a between-subjects
Fig. 3. Object switch condition. Percentages of time that 9-, 18-, and 27-month olds spent looking at the target objects during the final study trial and first test trial are shown. In this condition, two of the three objects in the display were replaced with novel ones between the study and test phases; the target objects were those that were replaced. The dotted line represents chance performance of 67%. Looking that is equally distributed among the three positions would result in infants spending 67% of the time fixating the objects in the switched locations. Preferences significantly greater than chance are marked with an asterisk.
86
J.L. Richmond et al. / Journal of Experimental Child Psychology 130 (2015) 79–91
variable showed a significant main effect of group, F(2, 43) = 16.71, p < .01, g2p = .44, but no effect of condition, F(1, 43) = 0.11, p > .05, and no Condition Group interaction, F(2, 43) = 0.46, p > .05. Independent-samples t tests confirmed that, averaged across conditions, 18-month olds (M = 25.7 s, SD = 4.6) accumulated a significantly greater amount of looking time during familiarization than did both 9-month olds (M = 14.5 s, SD = 4.3), t(30) = 7.10, p < .01, and 27-month olds (M = 18.6 s, SD = 7.4), t(28) = 3.18, p < .01. There was no difference in the amount of familiarization time accumulated by 9- and 27-month olds, t(28) = 1.89, p > .05. Experiment 1 showed that eye tracking may be a useful index of infants’ ability to bind objects to spatial locations. Our results showed that 18- and 27-month olds are sensitive to changes in the relative location of objects and look preferentially at objects for which the spatial relations have been manipulated. Although not statistically significant, there was also a trend for 9-month olds to look longer at the manipulated objects on location switch trials.
Experiment 2 Although there were no differences in study time as a function of condition, using a fixed trial length procedure resulted differences in the amount of familiarization time accumulated as a function of age. The fact that 9- and 27-month olds accumulated similar amounts of familiarization time, whereas only 27-month olds exhibited significant preferences for the objects that had switched locations, is consistent with what we know about age-related changes in encoding. As infants get older, information processing speed improves and they require less familiarization time to exhibit novelty preferences in traditional VPC tasks (Rose et al., 2004). It seems that the same is true in this visual preference paradigm. It is possible that 9-month olds would exhibit significant preferences on the location switch task given an age-appropriate amount of familiarization time. It is also possible that infants in this age range require relatively more study time in order to detect changes in object identity than they do to detect changes in object location. To test this hypothesis, in Experiment 2 we tested additional 9-month olds on both location switch trials and object switch trials; however, we used an infant-controlled habituation procedure during familiarization. Habituation procedures allow the infant to determine the length of each study trial and the duration of the familiarization period. On each trial, stimuli are presented until the infant looks away. Over the course of habituation, the length of looking on each trial decreases and the infant is said to have met a habituation criterion once the length of three consecutive trials is half the length of the first three trials. Because we were particularly interested in whether infants are capable of detecting changes in spatial relations within the first year of life, in Experiment 2 we chose to test only 9-month olds. We predicted that, given appropriate levels of familiarization time, infants would exhibit significant visual preferences on both the location switch and object switch tasks.
Method Participants An additional 19 infants (11 girls and 8 boys) were tested within 1 week of their 9-month birthday (Mage = 274 days, SD = 4). Data from a total of 6 infants were excluded due to fussiness (n = 1), insufficient looking on test trials (n = 4), or technical issues (n = 1); the final sample included a total of 13
Table 1 Mean accumulated familiarization times (s) as a function of age group and condition.
9-month olds 18-month olds 27-month olds Note. Standard deviations are in parentheses.
Location switch condition
Object switch condition
14.6 (4.2) 25.8 (4.6) 17.9 (8.7)
14.4 (5.3) 25.5 (6.4) 19.3 (7.5)
J.L. Richmond et al. / Journal of Experimental Child Psychology 130 (2015) 79–91
87
infants. Parents were reimbursed for travel expenses, and infants chose a small toy to take home as a ‘‘thank you’’ for participating.
Apparatus and stimuli The same eye-tracking system, stimulus presentation software, and stimuli used in the first experiment were used in Experiment 2. An additional two stimulus triplets (matched for presence/absence of faces, color, and category) were constructed to allow for a total of 8 possible test blocks.
Procedure Infants were seated on their parent’s lap, and the eye tracker was calibrated as in Experiment 1. Infants were tested on a maximum of 8 test blocks; each test block included an infant-controlled number of study trials followed by two 10-s test trials. During study trials, an array of the three toys was presented and the experimenter watched via a webcam and coded whether the infant was looking using the computer mouse. Each trial continued until the infant looked away, at which time the experimenter terminated the trial and the attention-getter was presented. Once the experimenter had confirmed that the infant was looking again, the next trial commenced and the experimenter coded looking using the mouse. The E-Prime script calculated the length of each trial online and determined when each infant had reached the habituation criterion. The habituation criterion was met when the average length of three consecutive trials was half the average length of the first three trials. Once infants had reached the habituation criterion, the attention getter was again presented to ensure that infants were fixating the center of the screen before the test trials were presented. As in Experiment 1, on half of the blocks the relative location of two of the three toys was switched (location switch condition) on the first test trial and returned to the original configuration on the second test trial. For the other half of the blocks, two of the objects were replaced with novel ones (object switch condition). Test trials were each 10 s long. Infants completed an average of 5.38 of the possible 8 blocks.
Results Raw eye-tracking data were summarized as in Experiment 1, and the mean percentage of time that infants spent looking at the target objects was analyzed during study and test. Analysis of looking time during the habituation phase confirmed that there was no difference in the amount of familiarization time accumulated as a function of condition, t(12) = 0.60, p > .05. On average, infants accumulated 32.02 s of familiarization time on location switch blocks and 35.08 s of familiarization time on object switch blocks, more than twice as much as 9-month olds in Experiment 1 (Mlocation switch = 14.63 s, Mobject switch = 14.36 s). Fig. 4 illustrates the mean percentages of time that 9-month olds spent looking at the target positions during study and test phases on both location switch and object switch trials. As in Experiment 1, a repeated-measures ANOVA was used to test for differences in the percentage of time infants spent fixating the target objects as a function of trial phase (study or test).2 There was a trend toward a significant main effect of trial in the location switch task, F(1, 12) = 3.92, p = .07, and a significant main effect of trial in the object switch task, F(1, 12) = 5.34, p = .04. Again, one-sample t tests confirmed that infants distributed looking behavior equally among the left, right, and bottom positions during study in both the location switch blocks (M = 70), t(12) = 0.77, p > .05, and the object switch blocks (M = 70), t(12) = 0.92, p > .05. At test, however, infants looked preferentially at the objects that had switched locations (M = 80), t(12) = 2.37, p < .05, and at the novel objects (M = 82), t(12) = 3.35, p < .05. 2 Looking times on the final trial of the infant-controlled habituation phase were very short and sometimes involved a single area of interest. For this reason, here the mean percentage of time spent looking at the target objects across all study trials was used rather than only the final trial.
88
J.L. Richmond et al. / Journal of Experimental Child Psychology 130 (2015) 79–91
Fig. 4. Mean percentages of time that 9-month olds spent looking at the target objects during study and test trials as a function of test condition (location switch vs. object switch). As in Experiment 1, the dotted line represents chance performance of 67%. Preferences significantly greater than chance are marked with an asterisk.
General discussion Here we have shown that, given sufficient encoding time, infants as young as 9 months learn about and are sensitive to disruptions in the spatial relations among an array of objects. In Experiment 1, we familiarized 9-, 18-, and 27-month olds with arrays of three objects and tested them with trials in which the location of two objects were switched and trials in which two objects were replaced with novel ones. On the location switch trials, 18- and 27-month olds looked preferentially at the objects that had moved locations, whereas 9-month olds did not. On the object switch trials, infants in all age groups did not look preferentially at the objects that had been replaced with novel ones. In Experiment 2, we explored the possibility that 9-month olds failed to look preferentially at switched objects because they did not accumulate as much familiarization time as older infants. By replicating the procedure using an infant-controlled habituation task, we showed that 9-month olds are sensitive to both the spatial relations and individual identities of an array of objects. Previous work using spatial search tasks has shown that infants begin to exhibit evidence of spatial relational memory at 24 months of age (Bushnell et al., 1995; Ribordy et al., 2013; Sluzenski et al., 2004). Why do young infants perform poorly on search tasks but exhibit spatial relational memory when tested on a looking time task? These data suggest that infants are able to encode information about the spatial relations among an array of objects and detect when these relations have been disrupted; however, it is clear that young infants are unable to use such representations to guide search behavior (Bushnell et al., 1995; Ribordy et al., 2013; Sluzenski et al., 2004). It is possible that the motoric demands of search tasks introduce additional processes that make it difficult for infants to express learning about spatial relations. Behavioral tasks likely recruit additional neural circuitry beyond the hippocampus; the relative immaturity of frontal circuits, for example, may limit the extent to which infants are able to express learning when tested on tasks involving complex motor behaviors. Alternatively, it is possible that differences in the allocentric versus egocentric demands of visual preference and spatial search tasks may play a role here. Research has shown that infants’ ability to guide spatial search behavior relative to the position of their own body (i.e., egocentric) develops earlier than the ability to guide search behavior based on the relative location of a set of cues (i.e., allocentric) (Vasilyeva & Lourenco, 2012). Although orientation using landmarks begins to emerge within the first year (Acredolo & Evans, 1980; Lew, Bremner, & Lefkovitch, 2000), the ability to use the relations between multiple distal cues is not seen until at least 24 months of age (Newcombe, Huttenlocher, Drummey, & Wiley, 1998; Ribordy et al., 2013). Blue et al. (2013) argued that visual
J.L. Richmond et al. / Journal of Experimental Child Psychology 130 (2015) 79–91
89
preference-based tasks, such as the one used here, tap allocentric processes because they require that the participant remember the location of objects relative to others in the environment. It could be argued, however, that such tasks are more akin to egocentric tasks because they test the ability to detect relational disruptions to an array of two-dimensional objects that are positioned in a fixed location on a screen in front of the infant. Infants likely use quite different processes to detect disruptions to static displays such as these relative to those required to move themselves through space to search for objects that must be located by assessing their relative position to other objects (Hayne & Imuta, 2011; Ribordy et al., 2013; Sluzenski et al., 2004). Future research should determine whether performance on eye-tracking measures of spatial relational memory is related to the development of spatial search behavior. It seems that infants’ ability to encode the relative position of objects may develop along a similar time course as their ability to bind together arbitrarily paired stimuli (Richmond & Nelson, 2009; Richmond & Power, 2014). Using another eye-tracking paradigm in which infants learned about pairs of faces and backgrounds, we showed that infants as young as 6 months will look preferentially at faces that were previously paired with the test background within a few hundred milliseconds of stimulus onset (Richmond & Nelson, 2009; Richmond & Power, 2014). In combination, these data suggest that infants are able to learn about arbitrary stimulus relations as well as the spatial relations much earlier than was previously thought. This work has a number of implications for visual preference methodology. Here we have shown that it is no longer necessary to rely on traditional ‘‘paired comparison’’ tasks; rather, with the advent of high-resolution eye tracking, it is now possible to use changes in the distribution of infant looking within novel and familiar stimulus arrays as an index of discrimination. Our data highlight the importance of considering age-related differences in the speed of encoding and point to potential differences in the amount of time that is required to encode information about object identity versus spatial relations. In Experiment 1, infants in all age groups failed to exhibit preferential looking under conditions in which two of the familiarization objects were replaced with novel ones (object switch) despite the fact that older infants were sensitive to disruptions in the spatial relations among similar arrays. In contrast, in Experiment 2, we showed that when an infant-controlled habituation procedure was used, 9-month olds exhibited preferential looking in both the object switch and location switch conditions. Infants’ failure to exhibit evidence of memory for object identity in Experiment 1 is a puzzling result that deserves some discussion. It is possible that infants accumulated insufficient familiarization time to allow them to detect the novel objects in the test array. Although allowing for longer familiarization did improve 9-month olds’ performance in Experiment 2, we think that it is unlikely that infants simply failed to encode identity information in Experiment 1. In performing the location switch task, infants encoded information about the identity of the objects and were able to bind it with location information to form a relational representation. In this sense, the familiarization time was sufficient to allow infants to encode both identity and location information. It is more likely that infants encoded the identity of the objects and the relations between them; however, the nature of this procedure is such that infants are better able to express memory for spatial relations than memory for identity. The visual preference procedure used here differs from standard VPC tasks in a number of important ways. In a standard VPC task, infants are typically shown a pair of identical objects during familiarization, and then one of those objects is replaced with a novel one during the test. Theories of infant visual attention suggest that infants seek to match each test stimulus with the representation that was stored during familiarization (Sokolov, 1963). In the current object switch task, rather than comparing a single familiar stimulus and a single novel stimulus at test, infants were required to simultaneously compare three test objects (two novel and one familiar). The fact that additional familiarization time improved 9-month olds’ performance suggests that stronger representations may be required to support these kinds of simultaneous comparisons. There are a number of important differences between the object switch and location switch conditions that may make expressing memory for identity more difficult than expressing memory for location. In the location switch condition, all three objects from the familiarization phase are presented during the test, just in a different configuration. In contrast, in the object switch condition, two of the three objects are replaced with novel ones, essentially removing most of the available retrieval cues. We know that infants’ performance on other memory paradigms is particularly sensitive to
90
J.L. Richmond et al. / Journal of Experimental Child Psychology 130 (2015) 79–91
changes in stimulus cues and may be disrupted if retrieval cues differ sufficiently between study and test (Hayne, 2004). Similarly, it is possible that the test phase in the object switch condition serves as a kind of context change, placing greater demands on retrieval than the test phase in the location switch condition. Each stimulus array contains contextual cues (i.e., category, presence/absence of faces, and color) that are not directly relevant but that may serve as additional retrieval cues. The consistency of these cues across study and test is more severely disrupted in the object switch condition than in the location switch condition because novel objects in the object switch condition were no longer matched on those categories. Future research should address whether infants are able to express memory for object identity in this paradigm if the novel objects are chosen to keep contextual cues such as category and color consistent across study and test phases. In sum, our work has shown that, much like relational memory for arbitrarily paired stimuli (Richmond & Nelson, 2009; Richmond & Power, 2014), infants encode information about the relative position of objects in the environment much earlier than they are able to use that information to guide search behavior. Eye-tracking methods have much potential to contribute to understanding of relational learning early in life; however, our understanding of how infants come to use relational information to guide flexible behavior continues to be limited by our ability to design paradigms to measure infants’ behavior. Acknowledgment This project was funded by a Discovery Project Grant from the Australian Research Council to J. L. Richmond. References Acredolo, L. P., & Evans, D. (1980). Developmental changes in the effects of landmarks on infant spatial behavior. Developmental Psychology, 16, 312–318. Astur, R. S., Taylor, L. B., Mamelak, A. N., Philpott, L., & Sutherland, R. J. (2002). Humans with hippocampus damage display severe spatial memory impairments in a virtual Morris water task. Behavioural Brain Research, 132, 77–84. Bachevalier, J., & Nemanic, S. (2008). Memory for spatial location and object–place associations are differently processed by the hippocampal formation, parahippocampal areas TH/TF, and perirhinal cortex. Hippocampus, 18, 64–80. Banta Lavenex, P., Amaral, D. G., & Lavenex, P. (2006). Hippocampal lesion prevents spatial relational learning in adult macaque monkeys. Journal of Neuroscience, 26, 4546–4558. Blue, S. N., Kazama, A. M., & Bachevalier, J. (2013). Development of Memory for spatial locations and object/place associations in infant rhesus macaques with and without neonatal hippocampal lesions. Journal of the International Neuropsychological Society, 19, 1–12. Bushnell, E. W., McKenzie, B. E., Lawrence, D. A., & Connell, S. (1995). The spatial coding strategies of one-year-old infants in a locomotor search task. Child Development, 66, 937–958. Cohen, N. J., & Eichenbaum, H. (1993). Memory, amnesia, and the hippocampal system. Cambridge, MA: MIT Press. Eichenbaum, H. (2004). Hippocampus: Cognitive processes and neural representations that underlie declarative memory. Neuron, 44, 109–120. Eichenbaum, H., Otto, T., & Cohen, N. J. (1992). The hippocampus – What does it do? Behavioral and Neural Biology, 57, 2–36. Filimon, F. (2010). Human cortical control of hand movements: Parietofrontal networks for reaching, grasping, and pointing. The Neuroscientist, 16, 388–407. http://dx.doi.org/10.1177/1073858410375468. Good, M. A., Barnes, P., Staal, V., McGregor, A., & Honey, R. C. (2007). Context- but not familiarity-dependent forms of object recognition are impaired following excitotoxic hippocampal lesions in rats. Behavioral Neuroscience, 121, 218–223. Hayne, H. (2004). Infant memory development: Implications for childhood amnesia. Developmental Review, 24, 33–73. Hayne, H., & Imuta, K. (2011). Episodic memory in 3- and 4-year-old children. Developmental Psychobiology, 53, 317–322. Henke, K. (2010). A model for memory systems based on processing modes rather than consciousness. Nature Reviews Neuroscience, 11, 523–532. Konkel, A., Warren, D. E., Duff, M. C., & Tranel, D. N. (2008). Hippocampal amnesia impairs all manner of relational memory. Frontiers in Human Neuroscience, 2, 1–15. Lew, A. R., Bremner, J. G., & Lefkovitch, L. P. (2000). The development of relational landmark use in six- to twelve-month-old infants in a spatial orientation task. Child Development, 71, 1179–1190. Mushiake, H., Tanatsugu, Y., & Tanji, J. (1997). Neuronal activity in the ventral part of premotor cortex during target-reach movement is modulated by direction of gaze. Journal of Neurophysiology, 78, 567–571. Newcombe, N., Huttenlocher, J., Drummey, A. B., & Wiley, J. G. (1998). The development of spatial location coding: Place learning and dead reckoning in the second and third years. Cognitive Development, 13, 185–200. Quinn, P. C. (2002). Category representation in young infants. Current Directions in Psychological Science, 11, 66–70. Ribordy, F., Jabès, A., Lavenex, P. B., & Lavenex, P. (2013). Development of allocentric spatial memory abilities in children from 18 months to 5 years of age. Cognitive Psychology, 66, 1–29. Richmond, J., & Nelson, C. A. (2009). Relational memory during infancy: Evidence from eye tracking. Developmental Science, 12, 549–556.
J.L. Richmond et al. / Journal of Experimental Child Psychology 130 (2015) 79–91
91
Richmond, J., & Power, J. (2014). Age-related differences in memory expression during infancy: Using eye-tracking to measure relational memory in 6- and 12-month-olds. Developmental Psychobiology, 56, 1341–1351. Rose, S. A., Feldman, J. F., & Jankowski, J. J. (2004). Infant visual recognition memory. Developmental Review, 24, 74–100. Ryan, J. D., Althoff, R. R., Whitlow, S., & Cohen, N. J. (2000). Amnesia is a deficit in relational memory. Psychological Science, 11, 454–461. Schutte, A. R., Spencer, J. P., & Schöner, G. (2003). Testing the dynamic field theory: Working memory for locations becomes more spatially precise over development. Child Development, 74, 1393–1417. Sluzenski, J., Newcombe, N. S., & Kovacs, S. L. (2006). Binding, relational memory, and recall of naturalistic events: A developmental perspective. Journal of Experimental Psychology: Learning, Memory, and Cognition, 32, 89–100. Sluzenski, J., Newcombe, N. S., & Satlow, E. (2004). Knowing where things are in the second year of life: Implications for hippocampal development. Journal of Cognitive Neuroscience, 16, 1443–1451. Sokolov, E. (1963). Perception and the conditioned reflex. Oxford, UK: Pergamon. Vasilyeva, M., & Lourenco, S. F. (2012). Development of spatial cognition. Wiley Interdisciplinary Reviews: Cognitive Science, 3, 349–362.
Contents lists available at ScienceDirect
Journal of Experimental Child Psychology journal homepage: www.elsevier.com/locate/jecp
What goes where? Eye tracking reveals spatial relational memory during infancy Jenny L. Richmond ⇑, Jenna L. Zhao, Mary A. Burns School of Psychology, University of New South Wales, Sydney, NSW 2052, Australia
a r t i c l e
i n f o
Article history: Received 25 November 2013 Revised 23 September 2014
Keywords: Spatial relational memory Binding Infancy Eye tracking Preferential looking Episodic memory
a b s t r a c t Episodic memory involves binding components of an event (who, what, when, and where) into a relational representation. The ability to encode information about the relative locations of objects (i.e., spatial relational memory) is a key component of episodic memory. Here we used eye tracking to test whether infants and toddlers learn about the spatial relations among objects. In Experiment 1, 9-, 18-, and 27-month olds were familiarized with an array of three objects. Following familiarization, they saw test arrays in which two of the objects had been replaced with novel ones (object switch condition) and arrays in which two of the objects had switched positions (location switch condition). Both 18- and 27-month olds looked significantly longer than would be predicted by chance at the objects that had switched spatial locations; however, 9-month olds did not. In Experiment 2, we showed that, given sufficient familiarization time, 9-month olds were also capable of detecting disruptions to the spatial relations among an array of objects. These results have important implications for our understanding of spatial relational memory development. Ó 2014 Elsevier Inc. All rights reserved.
Introduction Memories for events that we experience typically include details of who was there, what happened, when it happened, and where it happened. Compositionality is a key function of episodic memory ⇑ Corresponding author. Fax: +61 2 9385 3641. E-mail address: [email protected] (J.L. Richmond). http://dx.doi.org/10.1016/j.jecp.2014.09.013 0022-0965/Ó 2014 Elsevier Inc. All rights reserved.
80
J.L. Richmond et al. / Journal of Experimental Child Psychology 130 (2015) 79–91
(Henke, 2010); event memories are composed of individual components of an experience that are linked to form a coherent event representation. Episodic memories are composed of networks of relational representations (Cohen & Eichenbaum, 1993; Eichenbaum, Otto, & Cohen, 1992); relational networks allow event memories to be rapidly retrieved when a component cue is activated. Individual event networks are also connected to other events that share features, allowing us to make inferences about memories that are only indirectly related. Studies of lesioned animals (Eichenbaum, 2004), amnesic patients (Ryan, Althoff, Whitlow, & Cohen, 2000), and human brain activation (Konkel, Warren, Duff, & Tranel, 2008) have shown that the hippocampus is critically involved in episodic memory functions generally and memory for the relations among event components specifically. Relational memory refers to the binding of associated event elements, including individual items (who/what) as well as spatial (where) and temporal (when) information. Spatial relational memory allows us to remember ‘‘what’’ happens ‘‘where.’’ Remembering where you parked your car in the garage, or where your passport is in the filing cabinet, requires you to encode the position of the target object in relation to other objects in the environment. This kind of relational binding is essential for successful navigation and has been shown to depend on the hippocampus (Astur, Taylor, Mamelak, Philpott, & Sutherland, 2002; Banta Lavenex, Amaral, & Lavenex, 2006). Research looking at episodic memory development in toddlers has shown that children as young as 3 years are able to encode the relative location of objects in the environment and use this information to guide search behavior. For example, Hayne and Imuta (2011) had children hide three toys in specific locations in different rooms in their house (e.g., Big Bird is hidden in the bedroom behind the curtain). Following a 5-min delay, children were asked to verbally recall the identity and location of the toys along with the order in which they were hidden. Children were also asked to search for each object, and behavioral recall was coded. The results showed that although 3-year olds recalled less information than 4-year olds when asked about where the toys were hidden, spatial relational memory did not differ between the two groups. Both 3- and 4-year olds were able to accurately find the toys when asked to search for them. Spatial relational memory has also been studied in infants as young as 12 months using object search tasks. In these tasks, children watch the experimenter hide a toy and are encouraged to search for the toy after a delay (Bushnell, McKenzie, Lawrence, & Connell, 1995; Sluzenski, Newcombe, & Satlow, 2004). The accuracy of search behavior or the distance between the search position and hiding position (i.e., search error) is used as an index of memory. Children as young as 12 months can accurately search for a single object, particularly when its location is signaled by a distinctive cue; however, they have difficulty in using the relative location of two objects or cues to guide search behavior. For example, Bushnell et al. (1995) showed that 12-month olds were able to find a toy when it was hidden under a distinctive cushion, but not if it was hidden under a cushion that was in a particular location relative to a distinctive cushion. Similarly, Sluzenski et al. (2004) had children watch the experimenter repeatedly bury two toys that were always hidden in the same relative position to each other in a sandbox. Following a fixed training period, the experimenter hid the two toys out of sight of children, and then the location of one toy was revealed. Children were encouraged to search for the second toy using the relative location of the first one. Sluzenski and colleagues showed that 2- and 3-year olds were able to use spatial relations to accurately guide search behavior in this task; however, more than half of 18-month olds were unable to even complete the training. The ability to use the relative position of several cues to guide search behavior improves late in the second year of life (Ribordy, Jabès, Lavenex, & Lavenex, 2013; Sluzenski et al., 2004) and continues to improve during early childhood (Schutte, Spencer, & Schöner, 2003). Using a paradigm in which children searched an arena of overturned cups for hidden rewards, Ribordy et al. (2013) showed that it was not until 43 months of age that children reliably found rewards when searching in a complex array of 18 possible search locations. When only 4 possible search locations were used, 18- to 23-month olds still failed to reliably locate the cup containing the reward. Although these data, along with those of Sluzenski et al. (2004), suggest that spatial relational memory might not be functional during infancy, it is possible that infants learn about the spatial relations among objects before they are able to use this information flexibly to guide successful search behavior. Research using novelty preference tasks to study spatial relational memory in animals may be useful in informing our understanding of human development in this area.
J.L. Richmond et al. / Journal of Experimental Child Psychology 130 (2015) 79–91
81
Much evidence for the role of the hippocampus in spatial relational memory has been accumulated from animal studies using novelty preference tasks. These tasks capitalize on the animal’s natural tendency to explore novelty and allow similar procedures and dependent variables to be used across studies of rats and primates. Spontaneous preference tasks, such as the novel object recognition (NOR) task, are used in rat models. During the familiarization period, the rat is placed in an arena and allowed to explore two identical objects. The rat is removed from the arena, and two new objects are added; one object is identical to that seen during familiarization, and one is novel. The rat is placed in the arena again, and exploration is recorded. Memory is inferred if the rat exhibits a novelty preference, spending a greater percentage of time exploring the novel object than would be predicted by chance. Although object recognition is not affected by hippocampal lesions, when memory for the relative position of objects is tested, hippocampal rats exhibit impairments (Good, Barnes, Staal, McGregor, & Honey, 2007). For example, Good et al. (2007) tested rats with hippocampal lesions on the standard NOR task, in which familiar objects are replaced with novel ones at test, along with a version in which the relative location of objects was manipulated during the test. In this version, rats were familiarized with four objects and at test the location of two of the objects was switched. In the standard task, both hippocampal and control rats showed novelty preferences, exploring the novel objects more than the familiar ones. When tested on the spatial relational version, control rats spent a greater percentage of time than would be predicted by chance exploring the familiar objects that had switched locations; however, hippocampal rats did not explore the objects that had been switched for any longer than the objects that remained in the same configuration. These results suggest that the hippocampus is not critical for object recognition per se but is involved in integrating identity and location information into a relational network. Hippocampal animals are unable to encode the relative position of objects or recognize that the spatial relations in an array of objects have been changed (Good et al., 2007). Novelty preference tasks, like the visual paired comparison (VPC) task, have also been adapted to study spatial relational memory in primates. Bachevalier and Nemanic (2008) showed that monkeys with hippocampal lesions exhibit novelty preferences when tested on the VPC task with arrays that include novel objects, but not when tested on arrays in which the spatial relations among a set of objects have been manipulated. In this study, monkeys were familiarized with arrays of five objects and tested with arrays in which two or more of those objects had been replaced with novel ones (object identity condition) or arrays in which two or more of those objects had shifted relative position (object-in-place condition). The results showed that monkeys with hippocampal lesions exhibited novelty preferences when tested with object arrays containing novel items but exhibited null preferences when tested with arrays in which the identity of the objects remained the same but their relative position had changed. Blue, Kazama, and Bachevalier (2013) recently looked at the development of spatial relational memory in monkeys tested on the same task at 8 months, at 18 months, and as adults (5–6 years). In control animals, novelty preferences on this task increased with age; however, animals that had received neonatal hippocampal lesions exhibited null preferences at all ages. The ability to learn about spatial relations among an array of objects has a relatively protracted developmental trajectory and depends on the integrity of the hippocampus in monkeys (Bachevalier & Nemanic, 2008; Blue et al., 2013).
Experiment 1 Although novelty preference measures have been used extensively to study memory in human infants (Rose, Feldman, & Jankowski, 2004) and the extent to which infants form categories of spatial relations (Quinn, 2002), to date novelty preference measures have not been adapted to study spatial relational memory during infancy. Here we aimed to test whether 9- to 27-month olds learn about spatial relations by capitalizing on the spatial resolution of eye tracking. We chose this age range to include the youngest infants who have been shown to exhibit other kinds of relational memory (Richmond & Nelson, 2009) and to span the age range in which spatial relational memory has been demonstrated in search tasks (Sluzenski, Newcombe, & Kovacs, 2006).
82
J.L. Richmond et al. / Journal of Experimental Child Psychology 130 (2015) 79–91
There are several advantages to using eye tracking in this context. First, eye tracking allows us to test for spatial relational memory in infants who are motorically incapable of performing search tasks (i.e., 9-month olds). Second, there is much research to suggest that eye movements and gaze direction are critical in guiding search behavior (Filimon, 2010; Mushiake, Tanatsugu, & Tanji, 1997). Thus, by measuring eye movements directly, we are able to measure memory processes without the challenge of controlling for age-related differences in gross motor behavior. The procedure is adapted from the VPC task used by Bachevalier and Nemanic (2008), Blue et al. (2013); however, children were tested with single-object arrays rather than pairs of novel and familiar arrays. Children were familiarized with displays of three objects in a triangular spatial configuration (top left, top right, and bottom middle). On test trials, the familiar array was replaced with a novel array. In the object switch condition, two objects from the familiar array were replaced with new ones to create a novel stimulus. In the location switch condition, two objects from the familiar array switched locations to create a novel stimulus. If infants are sensitive to both the identity of objects and their spatial relations, we predicted that infants would look longer at the novel objects within the array on object switch trials and would look longer at the objects in the switched positions on location switch trials relative to the amount of looking that objects in these positions attracted during study. Alternatively, if children are unable to represent the spatial relations among objects until after 24 months of age, we predicted that 27month olds would look longer at the objects that have switched positions but that 9- and 18-month olds would not. Method Participants A total of 66 children were recruited from a database of families who had expressed an interest in participating in research; all were tested within 1 week of their birthday (9, 18, or 27 months). The 9-month olds (n = 22) were on average 276 days old (SD = 6), the 18-month olds (n = 24) were on average 547 days old (SD = 8), and the 27-month olds (n = 20) were on average 821 days old (SD = 4). There were 30 girls and 36 boys. Data from a total of 20 children was excluded due to fussiness (n = 8), insufficient looking on test trials (n = 4), calibration issues (n = 5), or technical issues (n = 3). The final sample included data from 16 9-month olds, 16 18-month olds, and 14 27-month olds. Parents were reimbursed for their travel costs, and children chose a small gift to take home as a ‘‘thank you’’ for participating. Apparatus and stimuli Fixation was recorded using a Tobii XL60 eye tracker. The eye tracker is composed of a 22-inch screen with a set of infrared cameras embedded in the perimeter. These cameras automatically capture a corneal reflection and, once calibrated, will track where the infant is looking on the screen. E-Prime software (Psychology Software) was used to present the stimuli and capture eye-tracking data. The stimuli were color photographs of toys (180 230 pixels, 11.7 12.5 cm on screen). In each block, children were presented with a group of three toys that were matched on the presence or absence of faces, color, and category (i.e., block toys, tower toys, or animal toys). There were a total of six triplet sets constructed. Procedure Following consent procedures, children were seated on their parent’s lap in front of the Tobii monitor. The position of the monitor was adjusted until the cameras detected the eyes, and a series of looming balls appeared at each of the corners of the screen and in the middle in a 5-point calibration sequence. The accuracy of calibration was checked and repeated if necessary. There were a total of 6 test blocks, each containing four 10-s study trials and two 10-s test trials (see Fig. 1). An attention-getter was presented before each trial to ensure that the infant was fixating the middle of the screen and that a good track was obtained before initiating each trial. During the study trials, three toys were presented on the screen in the top left, top right, and bottom middle positions. The configuration of the toys was consistent across the four study trials.
J.L. Richmond et al. / Journal of Experimental Child Psychology 130 (2015) 79–91
83
Fig. 1. Each block consisted of four 10-s study trials and two 10-s test trials. In the location switch condition (left panel), the position of two of the objects was switched at test. In the object switch condition (right panel), two objects were replaced with novel ones. Manipulated positions are marked for illustration only; the red ovals did not appear during stimulus presentation. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
In the first test trial, the three toys were presented again; however, two of the three toys were manipulated. On half of the trials, the position of two of the toys was switched (location switch condition). On the other half of the trials, two of the toys were replaced with novel ones (object switch condition). The positions that were manipulated were counterbalanced across trials (left–right, right–bottom, and left–bottom). On the second test trial, the original spatial configuration or identity of objects was restored. Blocks of trials continued until the infant was no longer interested in watching the pictures. There were a total of 6 blocks of stimuli, and 9-, 18-, and 27-month olds completed on average 5.63, 4.44, and 4.71 blocks, respectively. Results and discussion Consistent with previous work using eye tracking to measure relational memory (Richmond & Nelson, 2009; Richmond & Power, 2014), infants needed to spend at least 2500 ms1 fixating the stimuli during the test trial and contribute data on both location switch and object switch trials in order to be included in analyses. Areas of interest (top left, top right, and bottom middle) were defined a priori using inline code in the E-Prime script. The raw eye-tracking data were subjected to a fixation filter that determined the length of each fixation and the corresponding area of interest. The filter defined a fixation as a period of time in which the eye position does not move more than 50 pixels for at least 200 ms. Individ1 In our previous work using eye tracking to measure relational memory in infants, we required that infants spend 50% of the presentation time fixating the stimuli. Here we used 10,000-ms test trials; however, most infants spent only half of that time fixating the stimuli (M = 4870 ms, SD = 1824). We acknowledge that the trial length was likely too long, and rather than excluding more than half of the infants, we set the minimum test look criterion to be consistent with our previous work using a similar paradigm (Richmond & Nelson, 2009; Richmond & Power, 2014).
84
J.L. Richmond et al. / Journal of Experimental Child Psychology 130 (2015) 79–91
ual fixations were then analyzed to determine the percentage of time spent fixating the target positions during the study trials and the percentage of time spent fixating the manipulated positions during test. If attention is distributed equally among the objects in the left, right and bottom positions, we would expect infants to fixate objects in each position 33% of the time on average. During study, we predicted that infants would devote no more than 67% of the time to the ‘‘to be manipulated’’ objects. During test, if infants detected the spatial and/or nonspatial manipulations, we predicted that they would look disproportionately at the manipulated positions, fixating these objects significantly longer than would be predicted by chance (i.e., >67%). Analyses of variance (ANOVAs) were conducted separately for each condition (location switch and object switch) in order to test for differences in preferential looking as a function of age and trial phase. This approach is consistent with that applied in the original study using this method with primates (Bachevalier & Nemanic, 2008). Location switch condition Fig. 2 illustrates the mean percentages of time that 9-, 18-, and 27-month olds spent fixating the target objects (i.e., those in the switch positions) during the study and test phases of the location switch blocks. Here we analyzed the percentage of time spent fixating the target objects during the last familiarization trial relative to the first test trial using a repeated-measures ANOVA with age (9-month olds, 18-month olds, or 27-month olds) and trial phase (study or test) as factors. Data from the second test trial are not included in the analysis because the trial order was not counterbalanced and, as such, it is difficult to know to what extent infants’ looking behavior during this trial was influenced by the preceding test trial. This analysis revealed a main effect of trial, F(1, 43) = 7.11, p < .05, g2p = .14, but no main effect of group or Group Trial interaction (ps > .05). Averaged across age groups, infants spent a significantly greater percentage of time fixating the target objects that had switched locations during the test (M = 75, SD = 13) than during the final familiarization trial (M = 67, SD = 17). Although the ANOVA can tell us whether there is a difference in looking behavior as a function of age and/or trial phase, it does not speak to whether infants’ visual preferences differed from what would be expected by chance. One-sample t tests confirmed that during the final familiarization trial, infants in all age groups spent approximately 67% of the available looking time fixating the target objects. These preferences did not differ significantly from chance in 9-month olds, t(15) = 1.34, p > .05, 18-month olds, t(15) = 0.86, p > .05, or 27-month olds, t(13) = 0.08, p > .05. In contrast, during
Fig. 2. Location switch condition. Mean percentages of time that 9-, 18-, and 27-month olds spent looking at the target objects during the final study trial and first test trial are shown. In this condition, two of the three objects in the display switched locations between the study and test phases; the target objects were those that switched position. The dotted line represents chance performance of 67%. Looking that is equally distributed among the three positions would result in infants spending 67% of the time fixating the objects in the switched locations. Preferences significantly greater than chance are marked with an asterisk.
J.L. Richmond et al. / Journal of Experimental Child Psychology 130 (2015) 79–91
85
the first test trial, 18-month olds (M = 74), t(15) = 3.41, p < .05, and 27-month olds (M = 77), t(13) = 2.31, p < .05, looked significantly longer at the stimuli that had switched positions than would be predicted by chance. There was also a trend for 9-month olds to look preferentially at switched objects, (M = 75), t(15) = 2.12, p = .051. Object switch condition Fig. 3 shows the mean percentages of time spent fixating objects in the target positions during study and test trials. A repeated-measures ANOVA with age (9-month olds, 18-month olds, or 27month olds) and trial phase (study or test) as factors revealed no main effects or interactions (ps > .05). Much like the location switch trials, children spent on average 67% of the time fixating the target positions during the final familiarization trial. One-sample t tests showed that these preferences did not differ from chance (i.e., 67%) in 9-month olds, t(15) = 1.74, p > .05, 18-month olds, t(15) = 0.78, p > .05, or 27-month olds, t(13) = 0.55, p > .05. In the object switch condition, objects from two locations were replaced with novel objects during the first test trial. Surprisingly, there was no evidence of preferential looking at these novel objects during the test. The 9-month olds (M = 68), t(15) = 0.19, p > .05, the 18-month olds (M = 73), t(15) = 1.57, p > .05, and the 27-month olds (M = 73), t(13) = 1.21, p > .05, displayed preferences that did not differ significantly from chance, suggesting that they were not sensitive to changes in object identity. Familiarization time There is much evidence to show that the magnitude of infants’ novelty preferences is affected by the amount of study time given (Rose et al., 2004). In Experiment 1, infants were shown study configurations for a total of 40 s in both the location switch and object switch conditions; however, there may have been individual differences in the amount of familiarization time that was accumulated. For this reason, we wanted to rule out the possibility that differences in visual preference across age and condition could be accounted for by differences in initial study time. Table 1 contains the mean looking time accumulated during the familiarization period as a function of condition and age group. A repeated-measures ANOVA with condition (location switch or object switch) as a within-subjects variable and age group (9-month olds, 18-month olds, or 27-month olds) as a between-subjects
Fig. 3. Object switch condition. Percentages of time that 9-, 18-, and 27-month olds spent looking at the target objects during the final study trial and first test trial are shown. In this condition, two of the three objects in the display were replaced with novel ones between the study and test phases; the target objects were those that were replaced. The dotted line represents chance performance of 67%. Looking that is equally distributed among the three positions would result in infants spending 67% of the time fixating the objects in the switched locations. Preferences significantly greater than chance are marked with an asterisk.
86
J.L. Richmond et al. / Journal of Experimental Child Psychology 130 (2015) 79–91
variable showed a significant main effect of group, F(2, 43) = 16.71, p < .01, g2p = .44, but no effect of condition, F(1, 43) = 0.11, p > .05, and no Condition Group interaction, F(2, 43) = 0.46, p > .05. Independent-samples t tests confirmed that, averaged across conditions, 18-month olds (M = 25.7 s, SD = 4.6) accumulated a significantly greater amount of looking time during familiarization than did both 9-month olds (M = 14.5 s, SD = 4.3), t(30) = 7.10, p < .01, and 27-month olds (M = 18.6 s, SD = 7.4), t(28) = 3.18, p < .01. There was no difference in the amount of familiarization time accumulated by 9- and 27-month olds, t(28) = 1.89, p > .05. Experiment 1 showed that eye tracking may be a useful index of infants’ ability to bind objects to spatial locations. Our results showed that 18- and 27-month olds are sensitive to changes in the relative location of objects and look preferentially at objects for which the spatial relations have been manipulated. Although not statistically significant, there was also a trend for 9-month olds to look longer at the manipulated objects on location switch trials.
Experiment 2 Although there were no differences in study time as a function of condition, using a fixed trial length procedure resulted differences in the amount of familiarization time accumulated as a function of age. The fact that 9- and 27-month olds accumulated similar amounts of familiarization time, whereas only 27-month olds exhibited significant preferences for the objects that had switched locations, is consistent with what we know about age-related changes in encoding. As infants get older, information processing speed improves and they require less familiarization time to exhibit novelty preferences in traditional VPC tasks (Rose et al., 2004). It seems that the same is true in this visual preference paradigm. It is possible that 9-month olds would exhibit significant preferences on the location switch task given an age-appropriate amount of familiarization time. It is also possible that infants in this age range require relatively more study time in order to detect changes in object identity than they do to detect changes in object location. To test this hypothesis, in Experiment 2 we tested additional 9-month olds on both location switch trials and object switch trials; however, we used an infant-controlled habituation procedure during familiarization. Habituation procedures allow the infant to determine the length of each study trial and the duration of the familiarization period. On each trial, stimuli are presented until the infant looks away. Over the course of habituation, the length of looking on each trial decreases and the infant is said to have met a habituation criterion once the length of three consecutive trials is half the length of the first three trials. Because we were particularly interested in whether infants are capable of detecting changes in spatial relations within the first year of life, in Experiment 2 we chose to test only 9-month olds. We predicted that, given appropriate levels of familiarization time, infants would exhibit significant visual preferences on both the location switch and object switch tasks.
Method Participants An additional 19 infants (11 girls and 8 boys) were tested within 1 week of their 9-month birthday (Mage = 274 days, SD = 4). Data from a total of 6 infants were excluded due to fussiness (n = 1), insufficient looking on test trials (n = 4), or technical issues (n = 1); the final sample included a total of 13
Table 1 Mean accumulated familiarization times (s) as a function of age group and condition.
9-month olds 18-month olds 27-month olds Note. Standard deviations are in parentheses.
Location switch condition
Object switch condition
14.6 (4.2) 25.8 (4.6) 17.9 (8.7)
14.4 (5.3) 25.5 (6.4) 19.3 (7.5)
J.L. Richmond et al. / Journal of Experimental Child Psychology 130 (2015) 79–91
87
infants. Parents were reimbursed for travel expenses, and infants chose a small toy to take home as a ‘‘thank you’’ for participating.
Apparatus and stimuli The same eye-tracking system, stimulus presentation software, and stimuli used in the first experiment were used in Experiment 2. An additional two stimulus triplets (matched for presence/absence of faces, color, and category) were constructed to allow for a total of 8 possible test blocks.
Procedure Infants were seated on their parent’s lap, and the eye tracker was calibrated as in Experiment 1. Infants were tested on a maximum of 8 test blocks; each test block included an infant-controlled number of study trials followed by two 10-s test trials. During study trials, an array of the three toys was presented and the experimenter watched via a webcam and coded whether the infant was looking using the computer mouse. Each trial continued until the infant looked away, at which time the experimenter terminated the trial and the attention-getter was presented. Once the experimenter had confirmed that the infant was looking again, the next trial commenced and the experimenter coded looking using the mouse. The E-Prime script calculated the length of each trial online and determined when each infant had reached the habituation criterion. The habituation criterion was met when the average length of three consecutive trials was half the average length of the first three trials. Once infants had reached the habituation criterion, the attention getter was again presented to ensure that infants were fixating the center of the screen before the test trials were presented. As in Experiment 1, on half of the blocks the relative location of two of the three toys was switched (location switch condition) on the first test trial and returned to the original configuration on the second test trial. For the other half of the blocks, two of the objects were replaced with novel ones (object switch condition). Test trials were each 10 s long. Infants completed an average of 5.38 of the possible 8 blocks.
Results Raw eye-tracking data were summarized as in Experiment 1, and the mean percentage of time that infants spent looking at the target objects was analyzed during study and test. Analysis of looking time during the habituation phase confirmed that there was no difference in the amount of familiarization time accumulated as a function of condition, t(12) = 0.60, p > .05. On average, infants accumulated 32.02 s of familiarization time on location switch blocks and 35.08 s of familiarization time on object switch blocks, more than twice as much as 9-month olds in Experiment 1 (Mlocation switch = 14.63 s, Mobject switch = 14.36 s). Fig. 4 illustrates the mean percentages of time that 9-month olds spent looking at the target positions during study and test phases on both location switch and object switch trials. As in Experiment 1, a repeated-measures ANOVA was used to test for differences in the percentage of time infants spent fixating the target objects as a function of trial phase (study or test).2 There was a trend toward a significant main effect of trial in the location switch task, F(1, 12) = 3.92, p = .07, and a significant main effect of trial in the object switch task, F(1, 12) = 5.34, p = .04. Again, one-sample t tests confirmed that infants distributed looking behavior equally among the left, right, and bottom positions during study in both the location switch blocks (M = 70), t(12) = 0.77, p > .05, and the object switch blocks (M = 70), t(12) = 0.92, p > .05. At test, however, infants looked preferentially at the objects that had switched locations (M = 80), t(12) = 2.37, p < .05, and at the novel objects (M = 82), t(12) = 3.35, p < .05. 2 Looking times on the final trial of the infant-controlled habituation phase were very short and sometimes involved a single area of interest. For this reason, here the mean percentage of time spent looking at the target objects across all study trials was used rather than only the final trial.
88
J.L. Richmond et al. / Journal of Experimental Child Psychology 130 (2015) 79–91
Fig. 4. Mean percentages of time that 9-month olds spent looking at the target objects during study and test trials as a function of test condition (location switch vs. object switch). As in Experiment 1, the dotted line represents chance performance of 67%. Preferences significantly greater than chance are marked with an asterisk.
General discussion Here we have shown that, given sufficient encoding time, infants as young as 9 months learn about and are sensitive to disruptions in the spatial relations among an array of objects. In Experiment 1, we familiarized 9-, 18-, and 27-month olds with arrays of three objects and tested them with trials in which the location of two objects were switched and trials in which two objects were replaced with novel ones. On the location switch trials, 18- and 27-month olds looked preferentially at the objects that had moved locations, whereas 9-month olds did not. On the object switch trials, infants in all age groups did not look preferentially at the objects that had been replaced with novel ones. In Experiment 2, we explored the possibility that 9-month olds failed to look preferentially at switched objects because they did not accumulate as much familiarization time as older infants. By replicating the procedure using an infant-controlled habituation task, we showed that 9-month olds are sensitive to both the spatial relations and individual identities of an array of objects. Previous work using spatial search tasks has shown that infants begin to exhibit evidence of spatial relational memory at 24 months of age (Bushnell et al., 1995; Ribordy et al., 2013; Sluzenski et al., 2004). Why do young infants perform poorly on search tasks but exhibit spatial relational memory when tested on a looking time task? These data suggest that infants are able to encode information about the spatial relations among an array of objects and detect when these relations have been disrupted; however, it is clear that young infants are unable to use such representations to guide search behavior (Bushnell et al., 1995; Ribordy et al., 2013; Sluzenski et al., 2004). It is possible that the motoric demands of search tasks introduce additional processes that make it difficult for infants to express learning about spatial relations. Behavioral tasks likely recruit additional neural circuitry beyond the hippocampus; the relative immaturity of frontal circuits, for example, may limit the extent to which infants are able to express learning when tested on tasks involving complex motor behaviors. Alternatively, it is possible that differences in the allocentric versus egocentric demands of visual preference and spatial search tasks may play a role here. Research has shown that infants’ ability to guide spatial search behavior relative to the position of their own body (i.e., egocentric) develops earlier than the ability to guide search behavior based on the relative location of a set of cues (i.e., allocentric) (Vasilyeva & Lourenco, 2012). Although orientation using landmarks begins to emerge within the first year (Acredolo & Evans, 1980; Lew, Bremner, & Lefkovitch, 2000), the ability to use the relations between multiple distal cues is not seen until at least 24 months of age (Newcombe, Huttenlocher, Drummey, & Wiley, 1998; Ribordy et al., 2013). Blue et al. (2013) argued that visual
J.L. Richmond et al. / Journal of Experimental Child Psychology 130 (2015) 79–91
89
preference-based tasks, such as the one used here, tap allocentric processes because they require that the participant remember the location of objects relative to others in the environment. It could be argued, however, that such tasks are more akin to egocentric tasks because they test the ability to detect relational disruptions to an array of two-dimensional objects that are positioned in a fixed location on a screen in front of the infant. Infants likely use quite different processes to detect disruptions to static displays such as these relative to those required to move themselves through space to search for objects that must be located by assessing their relative position to other objects (Hayne & Imuta, 2011; Ribordy et al., 2013; Sluzenski et al., 2004). Future research should determine whether performance on eye-tracking measures of spatial relational memory is related to the development of spatial search behavior. It seems that infants’ ability to encode the relative position of objects may develop along a similar time course as their ability to bind together arbitrarily paired stimuli (Richmond & Nelson, 2009; Richmond & Power, 2014). Using another eye-tracking paradigm in which infants learned about pairs of faces and backgrounds, we showed that infants as young as 6 months will look preferentially at faces that were previously paired with the test background within a few hundred milliseconds of stimulus onset (Richmond & Nelson, 2009; Richmond & Power, 2014). In combination, these data suggest that infants are able to learn about arbitrary stimulus relations as well as the spatial relations much earlier than was previously thought. This work has a number of implications for visual preference methodology. Here we have shown that it is no longer necessary to rely on traditional ‘‘paired comparison’’ tasks; rather, with the advent of high-resolution eye tracking, it is now possible to use changes in the distribution of infant looking within novel and familiar stimulus arrays as an index of discrimination. Our data highlight the importance of considering age-related differences in the speed of encoding and point to potential differences in the amount of time that is required to encode information about object identity versus spatial relations. In Experiment 1, infants in all age groups failed to exhibit preferential looking under conditions in which two of the familiarization objects were replaced with novel ones (object switch) despite the fact that older infants were sensitive to disruptions in the spatial relations among similar arrays. In contrast, in Experiment 2, we showed that when an infant-controlled habituation procedure was used, 9-month olds exhibited preferential looking in both the object switch and location switch conditions. Infants’ failure to exhibit evidence of memory for object identity in Experiment 1 is a puzzling result that deserves some discussion. It is possible that infants accumulated insufficient familiarization time to allow them to detect the novel objects in the test array. Although allowing for longer familiarization did improve 9-month olds’ performance in Experiment 2, we think that it is unlikely that infants simply failed to encode identity information in Experiment 1. In performing the location switch task, infants encoded information about the identity of the objects and were able to bind it with location information to form a relational representation. In this sense, the familiarization time was sufficient to allow infants to encode both identity and location information. It is more likely that infants encoded the identity of the objects and the relations between them; however, the nature of this procedure is such that infants are better able to express memory for spatial relations than memory for identity. The visual preference procedure used here differs from standard VPC tasks in a number of important ways. In a standard VPC task, infants are typically shown a pair of identical objects during familiarization, and then one of those objects is replaced with a novel one during the test. Theories of infant visual attention suggest that infants seek to match each test stimulus with the representation that was stored during familiarization (Sokolov, 1963). In the current object switch task, rather than comparing a single familiar stimulus and a single novel stimulus at test, infants were required to simultaneously compare three test objects (two novel and one familiar). The fact that additional familiarization time improved 9-month olds’ performance suggests that stronger representations may be required to support these kinds of simultaneous comparisons. There are a number of important differences between the object switch and location switch conditions that may make expressing memory for identity more difficult than expressing memory for location. In the location switch condition, all three objects from the familiarization phase are presented during the test, just in a different configuration. In contrast, in the object switch condition, two of the three objects are replaced with novel ones, essentially removing most of the available retrieval cues. We know that infants’ performance on other memory paradigms is particularly sensitive to
90
J.L. Richmond et al. / Journal of Experimental Child Psychology 130 (2015) 79–91
changes in stimulus cues and may be disrupted if retrieval cues differ sufficiently between study and test (Hayne, 2004). Similarly, it is possible that the test phase in the object switch condition serves as a kind of context change, placing greater demands on retrieval than the test phase in the location switch condition. Each stimulus array contains contextual cues (i.e., category, presence/absence of faces, and color) that are not directly relevant but that may serve as additional retrieval cues. The consistency of these cues across study and test is more severely disrupted in the object switch condition than in the location switch condition because novel objects in the object switch condition were no longer matched on those categories. Future research should address whether infants are able to express memory for object identity in this paradigm if the novel objects are chosen to keep contextual cues such as category and color consistent across study and test phases. In sum, our work has shown that, much like relational memory for arbitrarily paired stimuli (Richmond & Nelson, 2009; Richmond & Power, 2014), infants encode information about the relative position of objects in the environment much earlier than they are able to use that information to guide search behavior. Eye-tracking methods have much potential to contribute to understanding of relational learning early in life; however, our understanding of how infants come to use relational information to guide flexible behavior continues to be limited by our ability to design paradigms to measure infants’ behavior. Acknowledgment This project was funded by a Discovery Project Grant from the Australian Research Council to J. L. Richmond. References Acredolo, L. P., & Evans, D. (1980). Developmental changes in the effects of landmarks on infant spatial behavior. Developmental Psychology, 16, 312–318. Astur, R. S., Taylor, L. B., Mamelak, A. N., Philpott, L., & Sutherland, R. J. (2002). Humans with hippocampus damage display severe spatial memory impairments in a virtual Morris water task. Behavioural Brain Research, 132, 77–84. Bachevalier, J., & Nemanic, S. (2008). Memory for spatial location and object–place associations are differently processed by the hippocampal formation, parahippocampal areas TH/TF, and perirhinal cortex. Hippocampus, 18, 64–80. Banta Lavenex, P., Amaral, D. G., & Lavenex, P. (2006). Hippocampal lesion prevents spatial relational learning in adult macaque monkeys. Journal of Neuroscience, 26, 4546–4558. Blue, S. N., Kazama, A. M., & Bachevalier, J. (2013). Development of Memory for spatial locations and object/place associations in infant rhesus macaques with and without neonatal hippocampal lesions. Journal of the International Neuropsychological Society, 19, 1–12. Bushnell, E. W., McKenzie, B. E., Lawrence, D. A., & Connell, S. (1995). The spatial coding strategies of one-year-old infants in a locomotor search task. Child Development, 66, 937–958. Cohen, N. J., & Eichenbaum, H. (1993). Memory, amnesia, and the hippocampal system. Cambridge, MA: MIT Press. Eichenbaum, H. (2004). Hippocampus: Cognitive processes and neural representations that underlie declarative memory. Neuron, 44, 109–120. Eichenbaum, H., Otto, T., & Cohen, N. J. (1992). The hippocampus – What does it do? Behavioral and Neural Biology, 57, 2–36. Filimon, F. (2010). Human cortical control of hand movements: Parietofrontal networks for reaching, grasping, and pointing. The Neuroscientist, 16, 388–407. http://dx.doi.org/10.1177/1073858410375468. Good, M. A., Barnes, P., Staal, V., McGregor, A., & Honey, R. C. (2007). Context- but not familiarity-dependent forms of object recognition are impaired following excitotoxic hippocampal lesions in rats. Behavioral Neuroscience, 121, 218–223. Hayne, H. (2004). Infant memory development: Implications for childhood amnesia. Developmental Review, 24, 33–73. Hayne, H., & Imuta, K. (2011). Episodic memory in 3- and 4-year-old children. Developmental Psychobiology, 53, 317–322. Henke, K. (2010). A model for memory systems based on processing modes rather than consciousness. Nature Reviews Neuroscience, 11, 523–532. Konkel, A., Warren, D. E., Duff, M. C., & Tranel, D. N. (2008). Hippocampal amnesia impairs all manner of relational memory. Frontiers in Human Neuroscience, 2, 1–15. Lew, A. R., Bremner, J. G., & Lefkovitch, L. P. (2000). The development of relational landmark use in six- to twelve-month-old infants in a spatial orientation task. Child Development, 71, 1179–1190. Mushiake, H., Tanatsugu, Y., & Tanji, J. (1997). Neuronal activity in the ventral part of premotor cortex during target-reach movement is modulated by direction of gaze. Journal of Neurophysiology, 78, 567–571. Newcombe, N., Huttenlocher, J., Drummey, A. B., & Wiley, J. G. (1998). The development of spatial location coding: Place learning and dead reckoning in the second and third years. Cognitive Development, 13, 185–200. Quinn, P. C. (2002). Category representation in young infants. Current Directions in Psychological Science, 11, 66–70. Ribordy, F., Jabès, A., Lavenex, P. B., & Lavenex, P. (2013). Development of allocentric spatial memory abilities in children from 18 months to 5 years of age. Cognitive Psychology, 66, 1–29. Richmond, J., & Nelson, C. A. (2009). Relational memory during infancy: Evidence from eye tracking. Developmental Science, 12, 549–556.
J.L. Richmond et al. / Journal of Experimental Child Psychology 130 (2015) 79–91
91
Richmond, J., & Power, J. (2014). Age-related differences in memory expression during infancy: Using eye-tracking to measure relational memory in 6- and 12-month-olds. Developmental Psychobiology, 56, 1341–1351. Rose, S. A., Feldman, J. F., & Jankowski, J. J. (2004). Infant visual recognition memory. Developmental Review, 24, 74–100. Ryan, J. D., Althoff, R. R., Whitlow, S., & Cohen, N. J. (2000). Amnesia is a deficit in relational memory. Psychological Science, 11, 454–461. Schutte, A. R., Spencer, J. P., & Schöner, G. (2003). Testing the dynamic field theory: Working memory for locations becomes more spatially precise over development. Child Development, 74, 1393–1417. Sluzenski, J., Newcombe, N. S., & Kovacs, S. L. (2006). Binding, relational memory, and recall of naturalistic events: A developmental perspective. Journal of Experimental Psychology: Learning, Memory, and Cognition, 32, 89–100. Sluzenski, J., Newcombe, N. S., & Satlow, E. (2004). Knowing where things are in the second year of life: Implications for hippocampal development. Journal of Cognitive Neuroscience, 16, 1443–1451. Sokolov, E. (1963). Perception and the conditioned reflex. Oxford, UK: Pergamon. Vasilyeva, M., & Lourenco, S. F. (2012). Development of spatial cognition. Wiley Interdisciplinary Reviews: Cognitive Science, 3, 349–362.
