TABLE 1.

Correlations

	CCI	LIN	AVG	DIV
CCI	0.99	0.98	0.97	0.98
LIN		0.99	0.97	0.98
AVG			0.98	0.97
DIV				0.99

Correlations between the discriminant weights used to read-out populations implementing different clutter rules. The diagonal in the table is the correlation of the weights vectors for the same populations obtained across different simulation runs, thus the values on the diagonal is an estimate of the upper-bound on the correlation values given the noise. CCI, complete clutter invariant; LIN, linear; AVG, average; DIV, divisive normalization.