実装レベルとしても神経科学的‧⽣物学的妥当性にも課題 ▪ ニューラルネットのニューロン数は⽣物の神経系に⽐べ⼤幅に少なく, ⼈よりも遥かに少ない⾍やネズミの 「数‧時間‧場所」に関する記憶メカニズムですら説明できない [Gallistel & King’09] 13 01 1. Computational theory: What is the goal of the computation, why is it appropriate, and what is the logic of the strategy by which it can be carried out? 2. Representation and Algorithm: How can this computational theory be carried out? In particular, what is the representation for the input and output, and what is the algorithm for the transformation? 3. Hardware Implementation: How can the representation and algorithm be realized physically? [Embick & Poeppel’15] Marr’s three levels 計算心理言語学の概観
実装レベルとしても神経科学的‧⽣態学的妥当性にも課題 ▪ ニューラルネットのニューロン数は⽣物の神経系に⽐べ⼤幅に少なく, ⼈よりも遥かに少ない⾍やネズミの 「数‧時間‧場所」に関する記憶メカニズムですら説明できない [Gallistel & King’09] 14 01 1. Computational theory: What is the goal of the computation, why is it appropriate, and what is the logic of the strategy by which it can be carried out? 2. Representation and Algorithm: How can this computational theory be carried out? In particular, what is the representation for the input and output, and what is the algorithm for the transformation? 3. Hardware Implementation: How can the representation and algorithm be realized physically? [Embick & Poeppel’15] Marr’s three levels ➢ 説明可能性や⽣態学的妥当性を慎重に評価する必要がある 計算心理言語学の概観
1. Computational theory: What is the goal of the computation, why is it appropriate, and what is the logic of the strategy by which it can be carried out? 2. Representation and Algorithm: How can this computational theory be carried out? In particular, what is the representation for the input and output, and what is the algorithm for the transformation? 3. Hardware Implementation: How can the representation and algorithm be realized physically? [Embick & Poeppel’15] Marr’s three levels 臨界期仮説に基づく学習効率の高い言語モデル