The purpose of the present study was to investigate the function of self-reinforcement (SR) in a situation where correct responses were objectively identified.<BR>In Experiment I, reinforcing effect was examined in a recognition task situation which was able to determine correct responses beforehand. First, seventy- eight 6th graders were shown the following number of non-sense syllables in sequence, one by one, and only once: 5, 10 and 15. In turn, they were shown another set of syllable items and told to identify them with the former set of items. This recognition task was repeated six times.<BR>The experimental conditions were as follows:<BR>(a) SR₊ (positive self-reinforcement): In the process of identification, Ss marked a × on their answering sheet if they thought their identification was confidently correct.<BR>(b) SR_- (negative self-reinforcement): In the process of identification, Ss marked a x on their answering sheet if they thought their identification was not correct.<BR>(c) SR_+- (both positive and negative self-reinforcement): In the process of identification, Ss marked a circle or × on their answering sheet according to their confidence.<BR>(d) N (non self-reinforcement): Ss identified each item only.<BR>The main results were:<BR>1) In SR₊ and SR_+- conditions, when a positive SR followed the first preceding response, the frequency of the same response to two consecutive presentations of the same items (SR₊ M₊ response) was significantly high. Thus, positive reinforcing effect was established (see FIG. 1).<BR>2) In SR_- and SR_+- conditions, when a negative SR followed the first preceding response, the frequency of the different response to two consective presentations of the same items (SR_-M_- response) was significantly low. Thus, negative reinforcing effect was not established (see FIG. 1).<BR>3) In N condition, the frequency of the same response to two consecutive presentations of the same items (M₊ response) was relatively high (see FIG. 1).<BR>In Experiment II, the reinforcing effect of positive SR was examined more accurately. Ss were sixty 6- th graders. The task was the same as in the Experiment I except for the followings. There were 10 items to be identified with another set consisting of 10 items. And the identification was repeated ten times.<BR>The experimental conditions were as follows:<BR>(a) SR₊₊ (positive SR, high incentive level): In the process of identification, Ss pushed the button to turn a lamp on indicating correctness, if they thought their identification was confidently correct.<BR>(b) SR₊ (positive SR, low incentive level): In the process of identification, Ss pushed the button to turn a lamp on indicating correctness, if they thought their identification was confidently correct.<BR>(c) N (non SR): Ss only identified each item.<BR>The major finding s were:<BR>1) As Experiment I, SR₊₊ and SR₊ conditions demonstrated the positive reinforcing effect (see TABLE 3).<BR>2) There was no difference between (a) and (b) conditions with respect to the reinforcing effect according to the incentive level (see TABLE 3 and 4).<BR>3) In SR₊₊ and SR₊ conditions, the frequency of the same response (M₊ response) was relatively high (see TABLE 3) and the number of correct responses were relatively more stable (see FIG. 5 and TABLE 5) than in N condition.