A test of Lamarckian theory over 50 generations of rats
Initial studies conducted indicate that offspring inherit traits that were acquired by parents during the latter’s lifetime. In a study conducted by Wilfred Agar, Frank Drummond, Oscar Tiegs and Mary Gunson at the University of Melbourne, the data collected suggest that certain characteristics that are learned by the parents are transferred to the offspring. The study was motivated by and simulated after a popular experiment that was done by a popular Harvard psychologist William McDougall. The Lamarckian evolution defines the phenomenon where learning and evolution can interact in such a way that the fitness of the individual is altered without any effect on the individual’s genetic code. The experiment conducted by McDougall was aimed at observing the Lamarckian effect through inheritance of light phobia in rats.
McDougall’s experiment was conducted over thirty-two generations with the observation indicating a progressive reduction in the number of errors. The reduction in the number of errors was attributed to an inheritance of the effects of ancestral training throughout the thirty-two generations. The conclusion that the number of errors reduced because of inheritance was accepted because there was no alternative theory. The Melbourne experiment was conducted to justify the McDougall experiment so as to establish if there is sufficient evidence to support the hypothesis. Also, the McDougall experiment did not publish the complete details of the performance of rats which indicated that a similar study was to be conducted to compare the results.
The apparatus used in the experiment was a water tank that had three parallel passage ways that were connected at the end of the tank. The rats had to scramble from the water through either of two escape routes that had bulbs that were lit alternatively. A rat that was in the water had the alternative of escaping on either the dimly-lit passageway or through a brightly lit passage where it was subjected to a three-second shock. The rat always had to learn to escape through a dimly lit passage independent of whether this was to the left or the right-hand side. The number of errors made, that of using a brightly lit exit were measured until the rat learned to use the dimly lit escape always. A rat was considered to have learned if it used the dimly lit escape for twelve consecutive times. The sample consisted of the control and the experimental group where the control group. The control group was used for comparison with the experimental group so as to ascertain that the findings could be attributed to the training and not other external conditions.
The findings of the McDougall experiment are detailed below
The rats to be bred from one generation was selected randomly and the rats studied were those from successive generations of trained rats. The gradual decline in the mean number of errors was considered as good evidence in supporting the Lamarckian effect. The two outstanding faults in McDougall’s experiment were the failure to have a control group. The absence of a control group means that the reduction in the number of errors made could be as a result of other external factors other than learning. Secondly, the findings of the experiment did not contain complete details of the performance of his rats with just the arithmetic mean as a measure of the scores.
The Melbourne experiment aimed at 50 generations of rats that were sampled over 20 years and were descendants of a single pair of rats. The first pair was divided into two where one was trained and the other untrained. The trained rat became the ancestor of the trained line of rats while the untrained rat became the ancestor of the control group. In every generation, the desired number of rats was trained and allowed to mate so as to be parents of the next generation. In the control group, the rats were not trained but rather allowed to mate so as to be parents of the next generation. However, certain litters of the control group were trained so as to provide control for the experimental group. Trained controls were not used for breeding and consequently the number of controls remained equal to the number in the trained group. The variables used in the experiment were
Year of the article reporting the result
Generation of the rat
Trained = Rat with trained ancestors
Control = Rat without trained ancestors
As for the group, but in later generations were divided into subgroups.
Number of errors
The number of errors before the achievement of “learning” (12 consecutive successful trials) or until special training was provided.
0 = no special training; 1 = special training
The appropriate summary measure to be used in the Melbourne’s researcher data is the mean. The mean provides a good measure of central tendency. Also, the fact that the number of errors to be considered does not exceed 52 indicates that the standard deviation is small, which indicates that the values are not extremely dispersed. Consequently, the mean will not be affected by extreme values and as such it will provide a good measure of central tendency. The t-test is used to evaluate if there is a significant difference between the mean values of the control and the trained groups. It is expected that the mean number of errors for the trained group will be significantly less than the mean number of errors for the control group. The graph that will most likely be appropriate for the data is the line graph. The average number of errors are being compared over the years and as such it is expected that the values will fluctuate. A line graph presents a better visual representation of the reduction in the mean number of errors encountered through the generation of rats.
Results for: rats_50.MTW
Descriptive Statistics: Number of errors
Variable Generation Mean StDev
Number of errors 1 72.2 46.2
2 50.45 36.10
3 63.90 48.54
4 54.95 44.15
5 53.21 38.29
6 41.16 39.21
7 43.75 37.22
8 46.65 43.69
9 34.14 30.23
10 47.40 39.06
11 30.88 24.85
12 40.87 39.81
13 43.86 49.20
14 16.39 12.78
15 26.52 24.54
16 24.15 27.35
17 27.28 32.55
18 42.09 41.70
19 22.70 22.74
20 31.60 37.80
21 26.88 32.34
22 24.95 29.95
23 24.08 24.17
24 22.80 24.46
25 21.74 25.87
26 23.84 27.07
27 22.73 27.79
28 24.62 25.82
29 26.78 25.59
30 31.31 38.38
31 26.82 26.93
32 26.65 29.16
33 27.27 26.11
34 42.06 39.91
35 37.73 31.51
36 36.10 36.56
37 39.24 35.14
38 35.06 30.79
39 27.75 21.79
40 26.77 19.61
41 24.65 19.80
42 27.30 28.20
43 26.93 29.16
44 23.43 29.42
45 18.49 15.92
46 27.43 24.04
47 24.83 24.09
48 22.57 17.21
49 25.96 24.21
50 22.56 23.46
The mean number of errors can be seen to reduce drastically over time. A graphical display of the mean number of errors across the generations of rats will have to be plotted so as to have a clear observation of the trend in the reduction. A line graph is appropriate for evaluating the mean number of errors across the generations of rats.
The line graph of the mean number of errors indicates that the average number of errors do decrease down the generation of rats. The observation was made by the McDougall experiment, and it would be of interest to observe if the same trend is witnessed across the categories “control” and “trained” in the variable group.
Descriptive Statistics: Number of errors
Variable Group Mean StDev
Number of errors Control 31.186 33.143
Trained 27.456 27.327
An observation of the line graph for the variable mean number of errors across the categories of the variable group indicates the same trend. For both the control and the experimental sample, the mean number of errors does reduce over time. The line graph cannot be used to provide sufficient proof that the characteristics that were learned by parents are passed on to the offspring. An alternative test will have to be used to evaluate if there is a statistically significant difference between the mean numbers of errors across the categories of the variable group. The independent sample t-test provides an alternative way of determining whether the mean number of errors in the control group is significantly different from the mean number of errors in the trained group.
Two-Sample T-Test and CI: Number of errors, Group
Two-sample T for Number of errors
Group N Mean StDev SE Mean
Control 2251 31.2 33.1 0.70
Trained 2360 27.7 27.9 0.57
An observation of the mean number of errors across the categorical variable group indicates that the control mean slightly higher than the trained mean. However, an independent sample t-test will have to be conducted to evaluate if the difference is statistically significant.
Difference = μ (Control) — μ (Trained)
Estimate for difference: 3.468
95% CI for difference: (1.696, 5.241)
T-Test of difference = 0 (vs ≠): T-Value = 3.84 P-Value = 0.000 DF = 4400
An independent sample t-test indicates that the mean number of errors differ across the variable group. The p-value obtained is 0.000 which indicates that the test is significant at 0.05 level of significance. It is therefore concluded that the mean number of errors in the control group is higher than the mean number of errors in the trained group. There is strong evidence to suggest that the experiment supports Lamarckian theory. The mean number of errors in the trained group is slightly lower than the mean number of errors in the control group.
Even though there exists a statistically significant difference in the number of errors across the categories control and trained, the graphical display of the average number of errors for the category control suggests that the mean number of errors will reduce down the generations. The line graph of the mean number of errors, however, suggests that the average number of errors does reduce down the generations for both the control and the experimental group. The reduction in the number of errors for both categories is a very critical observation as it suggests that there is another factor in play other than learning that causes the number of errors to reduce down the generations. Consequently, this suggests that the traits learned by parents are not passed down to offspring and as such the Lamarckian effect is not supported.
It would be essential for the researchers to evaluate why this is the case and whether there are other factors that make the rats reduce the number of errors other than training. If there are other factors that are contributing to a reduction in the number of errors then, it would be difficult to conclude that learned traits are passed down the generations from parents to offspring.
McDougall and Melbourne
The main shortcoming of the McDougall experiment is the fact that it did not have a control group which could be used to determine whether learning is passed down from parents to offspring. However, the line graph indicates that the number of errors decreases over the generations of rats. The Melbourne experiment was conducted to include control variable so that inherited characteristics from parents can be isolated as the sole factor that leads to a reduction in the number of errors down the generations. The findings of the experiment indicate that the mean number of errors for the experimental group is less than the mean number of errors for the control group. Further research needs to be done to identify the factors that led to a reduction in the number of errors in the control group. Once these factors are identified, they should be controlled for so as to estimate if the traits of parents are passed on to the offspring without affecting the genetic coding of the offspring.
28(4), 321-334.British Journal of Psychology. General Section. McDougall, W. (1938) Fourth report on a Lamarckian experiment (Part 1).