Table 3.

Cutoff ages for age effects in humans

Brain region	Age model	Cutoff age	Reduced sample n	Reduced sample P
Human neocortical gray matter	Linear	58 (57)	53 (52)	0.078 (0.092)
	Quadratic	61 (27)	57 (16)	0.051 (0.101)
	Cubic	65 (–)	63 (–)	0.071 (–)
Human frontal lobe gray matter	Linear	57 (–)	52 (–)	0.058 (–)
	Quadratic	50 (–)	46 (–)	0.051 (–)
	Cubic	41 (–)	34 (–)	0.080 (–)
Human neocortical white matter	Linear	74 (74)	79 (79)	0.145 (0.125)
	Quadratic	79 (74)	83 (79)	0.057 (0.215)
	Cubic	79 (74)	83 (79)	0.056 (0.206)
Human frontal lobe white matter	Linear	73 (69)	77 (74)	0.057 (0.108)
	Quadratic	74 (74)	79 (79)	0.106 (0.083)
	Cubic	74 (74)	79 (79)	0.119 (0.090)
Human hippocampus	Linear	82 (82)	85 (85)	0.071 (0.066)
	Quadratic	88 (88)	86 (86)	0.090 (0.086)
	Cubic	82 (82)	85 (85)	0.159 (0.148)

When datapoints older than or equal to the cutoff ages are removed, age is no longer significantly associated with brain region size. Boldface values indicate the best-fit models identified in Table 1. Cutoff ages and associated P values are identified using the two methods described in Materials and Methods; values in parentheses are for the method using degrees-of-freedom based on the full human sample size (maximum age = 88, n = 87). A dash indicates that the age effect remains significant even at the smallest possible sample size for which model parameters could be calculated (n = 7, with one brain at age 22 and six brains at age 23) when the model is evaluated as if it had the full sample size of n = 87.