Correction to Table 1 of our paper "Linked regularities..." (Science, 1995)

Richard B. Darlington and Barbara L. Finlay

December 7, 1995

Several readers have called to our attention a noteworthy error in Table 1 of our paper "Linked regularities in the development and evolution of mammalian brains" (Science 268:1578-1584, 1995). In that table we inadvertantly gave the regression statistics that are found when logged individual brain parts are predicted from the first principal component of the logged parts, while stating in the table's legend that the predictor was logged raw brain size. These latter two variables correlate .9980 with each other, but differ substantially in mean and standard deviation, thus making the figures in our Table 1 completely wrong when taken individually. The situation is saved somewhat by the fact that the regression slopes are at least proportional to the correct values. Thus for instance when Table 1 states that the slopes predicting paleocortex and neocortex are respectively .249 and .445, you can at least tell from those numbers that the ratio of the two relevant regression slopes in our Figure 1 is 1.79.

The correct figures are given below, where "a" is the additive constant in a simple regression, "b" is the regression slope, and "r" is the correlation. In each case the natural log of the structure shown is predicted from the natural log of total brain size.

Corrected Values for Table 1
STRUCTUREabr
Medulla-0.365.716.991
Cerebellum-1.563.947.995
Mesencephalon-0.963.736.990
Diencephalon-2.239.954.997
Paleocortex-0.384.685.963
Schizocortex-2.273.805.985
Hippocampus-1.204.775.983
Neocortex-3.0791.238.995
Septum-2.875.771.988
Striatum-2.862.980.997
Olfactory bulb*-0.022.454.704
*Not in published Table 1

Go to Darlington home page