THE UPWARD BIAS IN MEASURES OF INFORMATION DERIVED FROM LIMITED DATA SAMPLES

Extracting information measures from limited experimental samples, suc h as those normally available when using data recorded in vivo from ma mmalian cortical neurons, is known to be plagued by a systematic error , which tends to bias the estimate upward. We calculate here the avera ge of the bias, under certain conditions, as an asymptotic expansion i n the inverse of the size of the data sample. The result agrees with n umerical simulations, and is applicable, as an additive correction ter m, to measurements obtained under such conditions. Moreover, we discus s the implications for measurements obtained through other usual proce dures.