Professional Documents
Culture Documents
Schulz 2002
Schulz 2002
Schulz 2002
www.elsevier.com/locate/jneumeth
Received 29 May 2002; received in revised form 22 July 2002; accepted 23 July 2002
Abstract
We developed the sliding window correlation procedure in order to examine populations for possible heterogeneity in the ways
two variables are related with each other. This procedure involves computing correlation coefficients (R ) for overlapping successive
segments of the covariate scores. The distribution of resulting R s reveals fluctuations in the degree and direction of R over the
sample of ranked scores. This procedure is applied to behavioral data of aged rats, which were rank-ordered according to water
maze performance, and correlated with open field exploration and conflict behavior in a light/dark chamber. Results revealed
correlation coefficients of varying magnitudes and opposing directions for different segments of the population, which were
obscured by overall correlation analysis. E.g. for the superior learners, the R s were highest between maze learning ability, increased
open field exploration and reduced anxiety in the conflict test, whereas for the intermediate learners the R s were highest for maze
learning ability related with reduced exploration and increased anxiety. Thus, the sliding window correlation distribution can be
applied in conjunction with overall correlation analysis to provide information about the potential presence and locations of
subgroups within a population, especially if overall correlation analysis does not yield significant results.
# 2002 Elsevier Science B.V. All rights reserved.
Keywords: Individual differences; Aging; Correlation analysis; Learning and memory; Water maze; Open field exploration; Conflict behavior; Light/
dark chamber
hippocampal long-term potentiation, open field explora- purposes at the outset, so that any differences in their
tory behavior, and levels of water maze learning protocols were intended. The animals were group-
correlated significantly with each other in aged animals housed with two to three rats per cage unless cage-
which exhibited superior water maze performance, but mates had died during upbringing. They had free access
not in those with inferior performance (Schulz et al., to food and water.
2002). Subgroups of superior and inferior learners also
showed differences in correlation with levels of corti- 2.2. Morris water maze
costerone and adrenocorticotropin hormone measured
after restraint stress (Topic et al., unpublished). 2.2.1. Apparatus
Although performing separate correlation analyses The maze was a circular 185 cm diam swimming pool
for restricted segments of a distribution of scores (e.g. made of black polyethylene and filled 30 cm deep with
superior and inferior learners) may pinpoint possible 209/1 8C water. A 18 cm diam black polyethylene
fluctuations in the magnitude and/or direction of R platform was submerged 1.5 cm under the water surface
across the score dimension, this approach is not level. It was randomly located in the center of one of
sufficiently systematic. If there is variation in R , we four equally large imaginary quadrants (target quad-
would like to know the exact locations of variation rant) for all rats, but was maintained in a constant
along the ranked score distribution. In this article, we position throughout training for each rat. A sequence of
will propose a simple procedure, namely the ‘sliding four equidistant starting points was randomly varied for
window correlation’, to systematically uncover such each test day, but was the same for all animals during a
variation if it exists. This procedure involves carrying given test day. Cues located near the maze, such as
out correlation analyses for partly overlapping succes- posters, a blackboard and cupboard, were available for
sive segments of rank-ordered scores and to plot the spatial orientation and highlighted by four 75 W bulb
resulting R s. Such a procedure was shown below to lights. Diffuse ceiling lighting provided additional
reveal not only the presence and the degree of hetero- illumination (ca. 6.5 lx at the surface of the pool).
geneity in co-variation but also the course of develop- Computer-based systems for behavioral analysis were
ment of such interactions across samples of ranked placed behind a barrier remote from the maze. A camera
scores. To illustrate such results, we rank-ordered the (SONY, CCD-IRIS) was mounted 2 m above the maze
performance of aged rats in a spatial water maze and all trials were videotaped.
learning task and correlated this performance with
exploratory behavior in an open field. We also corre- 2.2.2. Procedure
lated water maze learning with performance in a light/ The rats were brought to a waiting room at least 30
dark choice chamber, a test for anxiety and exploration. min before the experiments and were kept in holding
cages shortly before testing. At the beginning of each
trial, they were individually placed into the pool facing
2. Materials and methods the wall. One adaptation trial of 2 min was run without
a platform present and was followed by 9 days of
2.1. Subjects training with a platform hidden. A training trial was
terminated when the animal escaped onto the platform
Our experiments were carried out in accordance with or after a maximum of 2 min, after which the rat was
the German Law on the Protection of Animals and were guided to the platform by an experimenter. Once on the
approved by the state authority (Bezirksregierung platform, it was left there for 30 s. Between the trials the
Düsseldorf). In the first experiment, 31 male 28 /30 animals were kept in holding cages for 60 s. After four
months old Wistar rats (531.77 g9/12.01 S.E.M.) trials in experiment 1 and after two trials in experiment
(animal facilities, University of Düsseldorf) were ranked 2, the rats were dried with paper towels and heated by
in terms of their performance in the water maze as two 75 W light bulbs. In experiment 2, two training
defined by the mean time required to learn the task. In a trials were run in the morning and two in the afternoon
second experiment, 27 male 26 months old Wistar rats every third day. All swimming trials were recorded on
(553.70 g9/14.03 S.E.M.) were ranked the same way. video and analyzed via Ethovision (Noldus, Wagenin-
Only those animals were included that did not show gen, Netherlands). Testing took place between 9:00 and
obvious signs of physical weakness or illness. The 31 17:00 h.
animals of the first experiment were maintained on a 12
h light:12 h dark cycle (lights on at 07:00 h), whereas the 2.2.3. Behavioral analysis
27 animals of the second experiment were exposed to a Time to escape onto the hidden platform and distance
reversed cycle (lights on at 19:00 h) such that all traversed before reaching the platform in the water maze
observations were made during the animals’ active night were measured for each acquisition trial. Daily mean
cycle. Experiment 1 and 2 were conducted for different times and distances were found to correlate highly (P -
D. Schulz, J.P. Huston / Journal of Neuroscience Methods 121 (2002) 129 /137 131
values ranged between 0.000006 and 0.000001 in both white compartment (24.5/35/29 cm) was illuminated
experiments). Therefore, only the times required to by a 25 W light bulb positioned circa 50 cm above the
locate the platform over the course of 9 training days floor. A video camera (SONY, DCR-TRV20E) was
was taken as an index of learning. Higher as compared mounted 1 m above the apparatus; all sessions were
with lower mean escape times averaged over all days of videotaped. A broad spectrum noise generator emitting
training (total mean latencies) were used to rank the 60 dB provided masking noise.
overall level of performance.
Table 1
Sliding correlation applied to ranked water maze performance correlated with variable X
132 D. Schulz, J.P. Huston / Journal of Neuroscience Methods 121 (2002) 129 /137
3. Results
superior learners is related to reduced anxiety and to To employ smaller intervals may be useful to check for
higher levels of exploratory activity. For this segment of the influence of one or the other variant animal within
learners, we previously established a relationship in the the learning distribution, which one suspects to con-
same direction between water maze performance and tribute substantially to the magnitude of a given
open field exploratory activity (see Fig. 3A). coefficient. Also, the optimum interval size for the
Interestingly, between the superior and intermediate detection of peaks may vary and can be assessed, e.g.
segments of learning levels, a gradual shift in the by using smaller windows for correlation analyses. On
opposite direction of correlation coefficients occurred, the other hand, with smaller interval sizes, of say five
such that for intermediate learning levels, high R s were animals or less, a given R may also become less
observed between water maze learning ability and meaningful.
increased anxiety. Consistently, for this segment of By contrast, larger interval sizes may generally
learning levels, we previously obtained high R s, indicat- smooth out the distribution of sliding R s. Although
ing a relationship between maze learning ability and the general shape of this distribution is likely to be
reduced open field exploratory activity (Fig. 3A). retained with the use of larger intervals, smaller changes
Furthermore, in segments of the sliding correlation within this distribution may remain undetected. How-
distribution of the inferior learners, water maze perfor- ever, with huge sample sizes, larger intervals may be
mance was unrelated with anxious behavior, as it was recommended for ease of computation. In such cases,
inconsistently related with open field exploratory activ- certain segments of the rank-order population may be
ity (Fig. 3A). re-checked using smaller interval sizes.
Thus, once again it becomes clear as to why the
overall correlation analysis did not yield a significant 4.3. Frequency of an animal contributing to the sliding
relationship between the variables tested: for different correlation
segments of the correlation distribution the variables
were differentially related with each other, i.e. positively One of the potential limits appertaining to the sliding
and negatively; an overall correlation analysis which correlation procedure concerns the frequency that a
included all sample points of the learning distribution particular animal contributes to the value of a given
thus masked such differentiated relationships. correlation coefficient. Assuming an interval size of ten
animals, for instance, the first animal of the rank order
4.1.3. Open field exploration and conflict behavior in an will be used only once for correlation analysis, the
L/D choice chamber second animal will be used twice, the third animal three
We also applied a variation of the sliding correlation times and so on, up to the tenth animal, from which
procedure to open field exploratory activity and L/D point on all of the animals will be used ten times, again
choice chamber behavior, since the latter is based on a up to the last nine animals of the rank order which will
conflict between the motivation to explore and the be used, respectively, nine, eight, seven, etc. times. Thus,
anxiety-induced motivation to remain in the black potentially, the shape of the sliding R distribution may
compartment of the L/D chamber. Thus, we performed be affected by the frequency with which each animal’s
the sliding correlation procedure for these variables weight contributes to it. Assuming that, e.g. the first
basing the rank order on a third variable, i.e. water maze animal of the rank order, a relatively superior learner in
performance levels. Interestingly, for most intervals of the present study, adds strongly to a particular direction
the rank order distribution, the coefficients were in the and strength of a correlation coefficient, the exclusion of
negative range, such that less open field exploratory such an animal may shift the next R of the rank order to
activity was associated with higher levels of anxiety. By a large degree and induce a strong slope in the
contrast, for the inferior segment of learning levels, the correlation distribution. Had the animal been included
R s were mainly in the positive range. Thus, the sliding for correlation analysis more often, a given peak
correlation procedure was again able to detect hetero- (positive or negative) may have consisted of more R s
geneity in the relationship tested within the aged animal being similar to each other, thus building a wider peak.
population. Although the extent of such a peak cannot be corrected
for, one may nevertheless ensure that the general shape
4.2. Sliding correlations using variable interval sizes of the correlation distribution is not affected by this
shortcoming. Heretofore we propose to compute a
Initially we set our interval sizes to an arbitrary sliding correlation control by minimizing the frequency
number of ten animals to perform the sliding correlation that each animal is included for correlation analysis.
analyses, although, of course, other interval sizes could Thus, instead of sliding down the rank order by one
also be used. In general, the smaller the interval size, the animal at a time, one may choose to slide by five animals
more detailed the sliding correlation distribution will be at a time (holding the size of the interval constant), or by
(as an example see Fig. 5B as compared with Fig. 5C). the number of animals which constitute one interval. In
D. Schulz, J.P. Huston / Journal of Neuroscience Methods 121 (2002) 129 /137 137
the present study, this procedure was demonstrated (Fig. of the relationship between variables across a distribu-
4B) and yielded a similar topography of the correlation tion of ranked scores. We have shown that the degree as
distribution. well as direction of relationship between behavioral
measures can vary greatly over the range of ranked
4.4. The sliding correlation procedure as a descriptive scores providing a basis for hypotheses regarding the
tool existence of subgroups in terms of the relationship
between these variables. While we have demonstrated
The sliding correlation procedure represents a de- the utility of this procedure on behavioral data of aged
scriptive tool, which may be used in conjunction with rats in pointing to such subgroups, the method has
overall correlation analysis. Given the many coefficients potential application to any field of study involving the
which are being computed by the new procedure, it analysis of relationship between variables.
would be inappropriate without proper adjustment for
Type I error to talk about significance levels. In the
present article the magnitude of the R s can be con-
sidered to provide ‘measures of effect’. In conjunction Acknowledgements
with overall correlation analysis, the sliding correlation
procedure can provide additional information about the This work was supported by a grant from the German
heterogeneity within correlation distributions. Such National Science Foundation (Grant number: Hu306/
information may be especially useful when overall 11-3). We would like to thank Professor J. Krauth and
analyses do not yield significant results contrary to Jay-Shake Li for their advice.
hypothesis as well as for the generation of new
hypotheses. Even in case of significant overall correla-
tion coefficients, the sliding window procedure may
References
reveal segments of the ranked score distribution respon-
sible for the correlation and may uncover potential Baxter MG, Gallagher M. Neurobiological substrates of behavioral
locations of heterogeneity. The procedure can be applied decline: models and data analytic strategies for individual differ-
to any field of study employing a correlation design and ences in aging. Neurobiol Aging 1996;17:491 /5.
may also be used to examine the effects of treatments or Costall B, Jones BJ, Kelly ME, Naylor RJ, Tomkins DM. Exploration
subject variables on the shape of the distribution of R s. of mice in a black and white test box: validation as a model of
anxiety. Pharmacol Biochem Behav 1989;32:777 /85.
As one control procedure, we suggest to establish a Crawley J, Goodwin FK. Preliminary report of a simple animal
random distribution of scores with which to perform the behavior model for the anxiolytic effects of benzodiazepines.
sliding correlation which then can be compared with the Pharmacol Biochem Behav 1980;13:167 /70.
one obtained with the actual data. In doing so, we Gallagher M, Burwell R. Relationship of age-related decline across
illustrated the degree of scatter that may result from several behavioral domains. Neurobiol Aging 1989;10:691 /708.
Hasenöhrl RU, Söderström S, Mohammed AH, Ebendal T, Huston
using randomized control scores. It was interesting to JP. Reciprocal changes in expression of mRNA for nerve growth
observe that even with randomized scores relatively factor and its receptors TrkA and LNGFR in brain of aged rats in
small peaks could be observed; however, the more relation to maze learning deficits. Exp Brain Res 1997;114:205 /13.
gradual change in magnitude and direction of the Issa AM, Rowe W, Gauthier S, Meaney MJ. Hypothalamic-pituitary-
coefficients over a wide range of ranks, as observed adrenal activity in aged, cognitively impaired and cognitively
unimpaired rats. J Neurosci 1990;10:3247 /54.
with the actual data, was absent in the control condi- Rapp PR, Amaral DG. Individual differences in the cognitive and
tion. neurobiological consequences of normal aging. Trends Neurosci
1992;15:340 /5.
Rowe WB, Spreekmeester E, Meaney MJ, Quirion R, Rochford J.
Reactivity to novelty in cognitively-impaired and cognitively-
5. Conclusion
unimpaired aged rats and young rats. Neuroscience 1998;83:669 /
80.
The sliding correlation procedure can uncover hetero- Schulz D, Huston JP, Jezek K, Haas HL, Roth-Härer A, Selbach O,
geneous or non-monotonic correlation distributions Luhmann HJ. Water maze performance, exploratory activity,
indicative of the presence of distinct subgroups within inhibitory avoidance and hippocampal plasticity in aged superior
a set of rank-ordered scores. As a complementary and inferior learners. Eur J Neurosci 2002, in press.
Sykova E, Mazel T, Hasenöhrl RU, Harvey AR, Simonova Z,
method to overall correlation analysis, the sliding Mulders WHAM, Huston JP. Learning deficits in aged rats
window correlation procedure provides a means of correlate with decrease in extracellular volume and loss of diffusion
visualizing the degree of heterogeneity or homogeneity anisotropy in hippocampus. Hippocampus 2002;12:269 /79.