|Home | About | Journals | Submit | Contact Us | Français|
The hippocampus is a brain region that is particularly affected by age-related morphological changes. It is generally assumed that a loss in hippocampal volume results in functional deficits that contribute to age-related cognitive decline. In a combined cross-sectional behavioural and magnetoencephalography (MEG) study we investigated whether hippocampal-associated neural current flow during a transverse patterning task - which requires learning relational associations between stimuli - correlates with age and whether it is modulated by cognitive competence.
Better performance in several tests of verbal memory, verbal fluency and executive function was indeed associated with higher hippocampal neural activity. Age, however, was not related to the strength of hippocampal neural activity: elderly participants responded slower than younger individuals but on average produced the same neural mass activity.
Our results suggest that in non-pathological aging, hippocampal neural activity does not decrease with age but is rather related to cognitive competence.
Normal aging is accompanied by a multitude of morphological changes in the brain. Consistently, volumetric MRI studies show a decline in total brain volume due to aging [e.g. [1-7]]. Although brain shrinkage is relatively widespread, brain regions differ in their intensity of volume loss [e.g. [8,9]]. Longitudinal [e.g. [8,10]], as well as cross-sectional studies [e.g. ] demonstrate that especially the caudate, the cerebellum, tertiary association cortices and the hippocampus are affected by shrinkage. Furthermore, Raz et al.  report that particularly in the hippocampus - essential for declarative memory, spatial navigation and contextual relations [12,13] - the magnitude of shrinkage accelerates with increasing age: the hippocampus shrinkage rate in healthy older people is more than twice that of younger people [1,14-16]. Volume reductions within the hippocampus were also analyzed in a recent MRI study with volume reduction found to be more severe in posterior regions . In sum, these observations demonstrate that the hippocampus is structurally especially affected by the aging process. The loss may be pronounced in pathological ageing, e.g., in Morbus Alzheimer the hippocampus is one of the regions being particularly vulnerable to degeneration .
Morphological changes, however, are not linearly related to brain function. Nonetheless, it is plausible to assume that regional brain damage will alter neural firing patterns in space and time. Hence, age-related changes in neural activation during memory tasks should be expected in the medial temporal lobe, notably the hippocampal region. On this account, several studies employing functional neuroimaging methods examined age effects on neuronal activation patterns during diverse memory tasks [e.g. [19-30]]. Results concerning age effects on activity of the hippocampal region differ across studies. Some studies report more activity in elderly people in right parahippocampal regions during memory retrieval , in the right hippocampus during autobiographical memory tasks  as well as higher cerebral blood flow increase in the medial temporal lobe during memory encoding . In contrast, there are also studies that report less activation for elderly people in mediotemporal structures, including the hippocampus and the parahippocampus, during encoding [21,24], encoding and retrieval , correct word recognition , working memory tasks [25,26] as well as during visual attention and episodic retrieval . Furthermore, there are also a number of studies that do not find age-related differences in the magnitude of activation in the hippocampal region during memory encoding [22,23,28,31-33].
Given that the hippocampus is thought to be responsible for the organization and structuring of memory storage [34,35] rather than for keeping the code of memory contents, this might explain why the results of studies concerning age-related differences in mediotemporal activation patterns during memory tasks are rather inconsistent. In the present study we therefore employed the transverse patterning task [36,37], i.e., contextual processing which consistently engages hippocampal function.
The transverse patterning paradigm requires learning nontransitive binary relations between three stimuli, pairs of which are presented. The well-known game "rock-paper-scissors" is an example of a transverse patterning paradigm with meaningful stimuli. The problem cannot be solved with elemental discriminations of single stimuli but requires learning relational associations [37,38]. The hippocampal formation is thought to play a central role in configural learning [12,13]. Consistent with this theory are the findings that damage of the hippocampal formation in humans [39-41], rats  and monkeys [42-44] impairs performance on the transverse patterning task. Furthermore, Driscoll and colleagues  found age-related differences in behavioral performance on a transverse patterning task: compared to younger people, elderly people were less often able to complete the transverse patterning task successfully, required more trials to reach criterion and made more mistakes. Here, we used magnetoencephalography (MEG) to track the fast neural responses induced by the transverse patterning task. A series of studies has demonstrated that activity from the medial temporal lobe can be assessed with MEG [46-63].
We specifically planned on assessing whether the hippocampal-associated activation pattern in healthy people changes with age. Furthermore, we recorded the extent to which this neural activity is associated with cognitive competence.
Fifty-two right-handed and healthy subjects (23 males and 29 females) ranging in age from 18 to 89 years (m = 52.4 years, sd = 19.6) (Additional file 1) with a mean MMSE score of 29.4 (SD = 1.08) participated in this study. Their mean education was 15 years (ranging from 10 to 22 years). None of the included participants reported a history of drug and/or alcohol abuse. Subjects were recruited by notifications posted on the campus of the University of Konstanz and in several residential homes for the elderly in the Konstanz area and by advertisement in the local newspaper and radio station. They were paid 30€ for their participation. Exclusion criteria were acute as well as a history of psychiatric diseases, a history of psychopharmacological medication, left-handedness (to control for potential influences of brain lateralization due to handedness), metal objects in the body as well as a history of severe head injuries or neurological problems (like epilepsy, strokes, brain tumors etc.). The ethics committee of the University of Konstanz approved this study.
Upon arrival at the laboratory, participants were familiarized with the MEG chamber, and study procedures and goals were clarified. All participants gave written informed consent. Afterwards subjects were screened with the MINI International Neuropsychiatric Interview  to assure that the participants did not suffer from a psychiatric disease. Subsequently, demographic data were assessed and handedness was determined using the Edinburgh Inventory . Furthermore, cognitive abilities were assessed with the CERAD-NP-plus test battery  with the subtests Verbal Fluency (VF = sum score of semantic and phonemic fluency), Word List Learning (WLL), Word List Delayed Recall (WLDR), Word List Recognition (WLR), Figure Recall (FR), Trail Making Test A and B (TMT-A/B). Additionally, the Digit Symbol and the Digit Span subtests of the German version of the Wechsler Adult Intelligence Scale [HAWIE-R; ] as well as the Benton Visual Retention Test - revised form  were conducted.
The two tasks and the basic design that were used in this experiment were adapted from the study of Hanlon and colleagues [50,51]: the Rock, Paper, Scissors task (RPS) and the Transverse Patterning task (TP) (Figure (Figure1).1). The well-known childhood game "Rock, Paper, Scissors" is an analogous task to the TP task with meaningful stimuli and relations. It was chosen as a training task to familiarize subjects with the underlying logic of transverse patterning designs.
Immediately prior to the MEG measurement, participants were trained on both tasks to familiarize themselves with the procedure. The first task to train was RPS where two out of three pictures of a hand symbolizing rock, paper or scissors were presented simultaneously. The significance of any one item depends on with which other item it is currently paired: When rock and scissors are paired, rock is correct, when scissors and paper are paired, scissors are correct, and if paper and rock are presented together, paper is correct (R+ S-, S+ P-, P+ R-).
In the second task (TP), the stimuli consisted of three disks in the colors blue, yellow or red. To solve the task, participants had to learn the following rule: When blue and yellow are presented together, blue is correct, when yellow and red are presented together, yellow is designated correct, and when red and blue are shown together, red is correct (B+ Y-, Y+ R-, R+ B-) (Figure (Figure1).1). Participants learned the abstract transverse patterning rule through explicit instruction.
In each trial of the RPS and TP task, two of the three pictures were presented next to each other on a grey computer screen using Presentation® (Neurobehavioral Systems, version 9.00). All stimuli were 9 × 10 cm (height × width). Participants were instructed to choose the correct picture as fast as possible by pressing the corresponding mouse button (left button for the left picture and right button for the right picture). The button press cleared the stimulus pair from the screen. Pictures were balanced for left and right side presentation.
Each trial began with the presentation of a grey screen with a fixation cross for 3-7 s to aid participants in maintaining gaze on the center of the screen, followed by the presentation of the stimulus pairs. Each pair of stimuli was presented for up to 10 s or until the participant responded via mouse button press. The response was followed immediately by a feedback sound: a high-pitched tone (1000 Hz) if the reaction was correct and a low-pitched tone (250 Hz) if the answer was incorrect. If the subject did not respond after 10 s, the trial was considered incorrect and the picture disappeared, the fixation cross appeared, and a new trial started (Figure (Figure2).2). Participants either needed to accomplish 10 correct training trials in a row or 42 trials altogether.
After training, MEG recordings were obtained, first for the RPS task to familiarize subjects with the MEG environment and then for the TP task. For artifact rejection, we recorded the electro-occulogram (EOG) using bipolar electrode configuration where electrodes were attached near the left and right outer canthus and below and above the right eye. For recording of the electrocardiogram (ECG), two additional electrodes were attached at the left lower forearm and the right collarbone. Subsequently, participants were seated in the magnetically shielded room (Vakuumschmelze Hanau) and their head shapes were digitized with a Polhemus 3 Space Fasttrack (Polhemus, Colchester, VT, USA). Five index points were determined to calculate the relative head position within the MEG sensor for source analysis. The subjects' head position relative to the pickup coils of the sensor was estimated before and after the measurement. During MEG measurement subjects were lying supine in a comfortable position. A video camera installed inside the chamber allowed monitoring subjects' behavior and ensured compliance throughout the experiment. To avoid artifacts, subjects were asked to stay still and to avoid head and eye movements as well as eye blinks during trials.
The experimental set-up and the stimuli were similar to the training procedure. Pictures were presented with a video projector (JVC™, DLA-G11E) with a refresh rate of 100 Hz on a white plastic screen attached to the ceiling of the room. Subjects responded by pressing the corresponding button on a button box with their right hand. In order to avoid magnetic interference, the feedback tones were presented through ear tubes. Sound intensity was adapted to the individual hearing level before measurement. MEG was recorded continuously and digitized at a rate of 678.17 Hz using a 148-channel whole head magnetometer (MAGNES™, 2500 WH, 4 D Neuroimaging, San Diego, USA). A real band-pass filter of 0.1 - 200 Hz was used for data acquisition. EOG and ECG were recorded with a SynAmps amplifier (Neuroscan ™) using Ag/AgC1 electrodes.
The experiment ended either after the subject had achieved 120 correct trials or when 180 trials in total had been presented.
Since the RPS task was chosen as a training task in order to familiarize subjects with the transverse patterning design and the MEG environment, data analysis was solely conducted for the transverse patterning task. After noise-reduction of environmental magnetic noise (noise reduction procedure, 4 D Neuroimaging), using the reference channels of the WH2500 data were analyzed using BESA (Brain Electrical Source Analyses; MEGIS Software GmbH) software version 5.2.4.. Cardiac activity and eye movements were removed with a semi-automated procedure implemented in BESA. For each subject, epochs with a 200 ms baseline and a post trigger (stimulus presentation) window of 800 ms were generated. After baseline correction, a bandpass-filter of 1-40 Hz (Low Cutoff filter: 1 Hz, forward, 6 dB/oct; High Cutoff filter: 40 Hz, zero phase, 24 dB/oct) was applied to the resulting data epochs. Afterwards, data epochs were averaged separately for correct and incorrect trials. In order to transform surface MEG into brain source activity, a source montage with 51 fixed regional dipoles in a homogeneous sphere was defined . In the case of MEG a regional dipole consists of two perpendicular, fixed equivalent current dipoles describing the two tangential components of a source, while the radial component does not contribute to the magnetic field outside the volume conductor. As we were interested in brain activity originating from hippocampal regions, the source montage consisted of 2 regional dipoles (Talairach coordinates: x = ± 42.4; y = -9.5; z = -23.9 for the right/left hemisphere) for assessing activity from the medial temporal lobes and 49 other regional dipoles that represented other brain areas, thus increasing the sensitivity of the fixed sources that had been positioned in the hippocampal region. All dipoles of the source montage were set by hand with fixed location and orientation in similar distance to the head surface to minimize differences in the sources' explanation of variance. In order to explore the time window for further analyses of dipole strength, a grand average of the correct trials of all subjects was generated. The time segment of 130-220 ms after stimulus presentation, containing the peak amplitude of the hippocampal source waveforms in both hemispheres, was chosen for further analyses (Figure (Figure33 and and4).4). After selecting the time window of interest, dipole analysis was carried out on each participant. By using the source montage in the individual headframe, the mean of the norm of the dipole moment of the hippocampal-associated sources was calculated for the above-mentioned time segment per person and then analyzed statistically.
Statistical analyses were conducted using R 2.8.1 [R Development Core ] and SPSS (SPSS Statistics 17.0). For assessing age effects on test results (cognitive capacity) and behavioral data (number of training trials needed to reach criterion, number of errors, reaction times), linear regressions were conducted with the constant variable age and dependent variables test results and behavioral data, respectively. For examining effects of age and cognitive competence on hippocampal activity, repeated measurement analyses of covariance (ANCOVA) with the factor hemisphere (left, right) and the covariates age and neuropsychological test results were conducted. In order to capture possible nonlinearities, restricted cubic spline transformations of age were investigated , but they did not improve model fit as measured by AIC  and were discarded.
Significant effects of age on cognitive performance were found for several tests. Elderly people performed worse in the digit symbol test, the digit span test, Benton test, Boston naming test, Mini Mental State Examination, word list learning, word list delayed recall, figure recall, trail making test version A and version B. No significant effects of age were found for verbal fluency, figure drawing as well as word list recognition (for details see Table Table11).
There was no age effect on the number of training trials needed to reach criterion or on the number of incorrect trials during MEG measurement. Only reaction times were significantly longer for elderly people (Table (Table22).
For the selected time window of 130-220 ms after stimulus presentation, subjects showed a mean source strength of 2.18 nAm (SD = 1.19) for the left and 1.97 nAm (SD = .99) for the right hemisphere. Statistical analyses did not show significant effects of age (F(1,50) = .41; p = .53) (see Figure Figure5),5), sex (F(1,50) = .27; p = .61) or hemisphere (F(1,51) = 1.94; p = .17) on the mean amplitudes of the hippocampal sources.
There were significant relationships between several neuropsychological test results and hippocampal activation (see Table Table3).3). The influence of two verbal memory tests were positive, i.e. higher scores in word list learning (p = .03) and word list delayed recall (p = .04) were associated with higher hippocampal activity (β(WLL) = .08; β(WLDR) = .14). Additionally, higher scores in both versions of the trail making test (which means worse performance) were associated with lower hippocampal activity (p(TMT-A) = .03; β(TMT-A) = -.02; p(TMT-B) = .045; β(TMT-B)= -.01) (Figure (Figure66).
Furthermore, there were statistical trends of positive relationships between the digit symbol test and hippocampal activity (p = .08; β(Digit Symbol) = .02) and verbal fluency and hippocampal activity (p = .08; β(VF) = .03). The relationship between the covariates digit span test, Benton test, word list recognition as well as figure recall with hippocampal activity did not reach significance.
This study examined whether hippocampal activity during the transverse patterning task varies with age and whether it is also modulated by cognitive competence. Our study yielded two main findings: First, no age effect on hippocampal activity was observed. Second, hippocampal activity was associated with cognitive competence. The strengths and limitations of this study are summarized in table table44.
Young and elderly people showed a similar pattern of hippocampal source strength. This is in line with a number of studies that also did not find age-related differences in the extent of hippocampal activation during memory tasks [22,23,28,31-33,73]. These studies provide evidence of intact hippocampal activation during healthy aging, which is consistent with our findings. Many other studies assessing neuronal activity during memory tasks, however, found either more [20,29,30] or less [19,21,24-27] activity in the hippocampal region for elderly people. They typically investigated brain activation during memory tasks that activate a wide network of different brain areas and therefore it is not surprising that the results are inconsistent. In this study however, we used a hippocampus specific task and neither did we observe age-related changes in hippocampal current flow nor age-related differences in behavioral performance. At first glance this seems to be contradictory to the age-related deficits in behavioral performance found by Driscoll and colleagues . However, the lack of age-related differences in our study could be explained by differences in experimental setup: Driscoll et al. (2003) used a stepwise approach where subjects had to detect the rules on their own by trial and error which is most likely more difficult than learning the rules by explicit instructions. Given that older adults show longer periods of task acquisition and lower mastery than their younger counterparts with increasing task difficulty [e.g. ], it is not surprising that the performance of the elderly subjects was inferior to the performance of the young group in the Driscoll (2003) study. In our study, participants learned the rule by explicit instruction and except for the reaction times, we did not observe age-related deficits in the performance on the transverse patterning task. Since young and elderly performed similarly well, age-related differences in activation of hippocampal-associated components could not necessarily be expected.
Hippocampal activation seems to differ between healthy elderly people and people with mild cognitive impairment (MCI) or Alzheimer's disease (AD). While we did not find age-related differences in hippocampal activation in our healthy sample, it was reported that people with AD show decreased hippocampal activity [31,75], whereas people with MCI paradoxically show more hippocampal activation [32,75]. Less impaired people with MCI show a pattern of hippocampal hyperactivation, whereas more impaired persons demonstrate hippocampal hypoactivation . Miller and colleagues  found that greater hippocampal activation during memory tasks in people with MCI serves as a predictor for greater subsequent cognitive decline. Furthermore, they suggested that hippocampal hyperactivation may represent an attempted compensatory response to accumulating neurodegenerative pathology. In addition, Dickerson and colleagues  hypothesized that increased hippocampal activation in MCI may serve as a marker for impending clinical decline. In fact, plaques disrupt the synchrony of convergent inputs and thus may lead to an increased number of failures for signal transmission . This in turn may lead to larger rather than smaller PSPs and with it to a smaller intracellular current flow (which produces the MEG signal).
One further explanation for our results could be that we had a cognitively still highly fit sample: there was neither a significant effect of age on the number of training trials needed to reach criterion, nor on the number of errors made during MEG measurement. Furthermore, we merely investigated the correct trials. As could be expected, elderly people responded significantly slower than younger people but besides this, young and elderly people performed similarly well. Therefore, age-related changes in neuronal activation patterns could not necessarily be expected.
Our data reveal a significant relation between cognitive competence and hippocampal source strength: better performance in tests for verbal memory, executive functioning, verbal fluency, and cognitive speed was associated with higher hippocampal activity, whereas worse performance in executive functions (higher scores in the Trail Making test) was related to lower hippocampal activity. As mentioned above, several studies exist which examine age-related hippocampal activation during different experimental tasks and related performance in these tasks [e.g. [19-24,27-33,73]]. However, to our knowledge, this study is the first to examine the correlation between hippocampal activity and independently assessed general cognitive capacity. Results indicate that there might be a more general relationship between medial temporal lobe activity and cognitive performance, with higher general cognitive competence being associated with higher hippocampal activity. This is also in line with the findings from the MCI and AD studies mentioned above [31,32,75-77]: the more cognitively impaired the participants were in these studies, the lower was their hippocampal activation. Further studies are needed to clarify and confirm the significance of these findings.
Results indicate that MEG is a useful method to investigate functional changes in deeper brain structures such as the hippocampus, in line with several earlier studies using similar methods [e.g. [50,51,58,63]]. One limitation of MEG technology, however, is that the detected activity might not have originated from the hippocampus directly but from adjacent temporal lobe structures. However, the transverse patterning task is a well-known paradigm for assessing hippocampal function, which has been demonstrated in many studies (e.g. Astur 1998; Reed and Squire 1999; Hanlon, Weisend et al. 2003; Astur and Constable 2004; Hanlon, Weisend et al. 2005; Meltzer 2007; Moses 2007).
It is generally assumed that age-related morphological changes in the hippocampus result in functional deficits that contribute to cognitive decline. In the present study we employed the hippocampus-dependent transverse patterning task to assess whether hippocampal activity varies with increasing age and whether it is also associated with cognitive competence. To our knowledge, this study is the first to examine the correlation between hippocampal activity and general cognitive capacity in healthy aging. While one might assume that hippocampal activity would decrease with increasing age, we found no such effect in healthy aging. In addition, our results suggest that hippocampal neural activity during the transverse patterning task is associated with cognitive competence. In sum, our results support the notion that in non-pathological aging, hippocampal neural activity does not necessarily decrease with age but is rather related to cognitive competence.
VML, ITK and TE designed the experiment. VML collected the data and performed the MEG data analysis with support of CW and IPJ. SK performed the statistical analyses. VML drafted the manuscript and ITK, CW, SK, IPJ as well as TE revised it critically. All authors commented on and approved the final manuscript.
Sample description. The table illustrates each volunteer's age, sex and associated diseases.
This study was funded by a grant of the Heidelberg Academy of Sciences awarded to ITK, and a PhD fellowship of the German National Academic Foundation awarded to VML. The authors thank Johanna Fiess, Christiane Wolf and Ursula Lommen for assistance in data collection as well as Patrick Berg for support in MEG data analysis.