本文以犯罪學近300 年來從以對犯罪人理性選擇的理解開始，討論到過 去一百年犯罪學研究證據顯示犯罪人也有「無意識理性」（或稱「自動化思 考」）、生理病理、環境因素的多重成分，但是當前刑罰制度的設計，卻承繼 了三百年前的遺絮，認定多數犯罪人為理性行為或自由意志的選擇，未能根據 科學證據或典範轉移而改變。本文提出不犯罪與犯罪的理性光譜，認為刑罰或 司法能在此光譜的中間（功利理性、無意識理性）運作，難以在光譜兩極端 （道德理性與生理病理性）發揮效果。重新還原犯罪行為的魔術方塊，有助傳 統刑事司法的刑罰主義、責任主義，慢慢轉化成具有治療或矯正效果的治療式 司法或以預防主義為導向。刑法第12-24條部分條文以及第57條是科刑的重要 根據，但本文認為各項條文中，頗有違反或落後於科學證據之處。因此提出行 為人責任盤點表與科刑審酌Jou's 量刑處遇量表，作為具體化相關科刑條文的 工具。責任盤點表可由法院自行審酌分類，Jou's 量表則可由法院在量刑前委 託各種專業團隊調查完成，當中需包括精神醫學、心理、社工、犯罪學等專業 人員評估計分。為了初步檢驗Jou's 量表的可行性，本文以六件涉及死刑的重 大案件為例，排除兩案正在審理中外，根據量表分析結果，四案中至少一案已 經執行死刑案件有待商榷。本文最後提出未來量刑工具化的建議，提供司法院 或法院進行實驗，並給予經驗考驗。

The assumption of the rational criminal has been the cornerstone of modern penal systems worldwide for over three hundred years. Criminological research, however, has revealed much more complexity underlying crime and criminal behavior; distinguishing sociological, psychological, genetic and pathological issues. Yet, the modern penal system seemingly largely ignores such scientific findings, in its determination to retain rationality and thus its blame-allocation and punishment-orientation. This paper argues human behavior, both normal and deviant are controlled by moral, utilitarian and habitual rationality as well as non-rational forces such as genetic defect, neuropathy, mental disability and other pathological factors. Any penal reform needs to start with a more scientifically sound basis and understanding of the multiple origins of criminality. Sentencing for incaraceration (prison) works only for criminals with definite utilitarian motivation and sometimes for habitual criminals. But an alternat i ve therapeutic and treatment-oriented penal system is needed for most habitual and all pathological criminals. When it comes to sentencing or treatment decisions, Taiwan Criminal code §12-24 focuses on determining criminal responsibility, and 57 works to mainly adjust for mitigation and aggravation factors. Current provisions as drafted in Criminal Code § 57 lag some distance behind present scientific criminological evidence: indeed, irrelevant factors are often cited. Instead, this paper, proposes in any individual case, the adoption of Jou's Sentencing/Treatment Instrument to measure the harm of the crime-at-issue and the risk of future recidivism of the criminal. Use of such an instrument would assist in appropriate structuring of judicial discretion, taking account of the best scientific evidence. It is also proposed that a pre-sentencing report be prepared and the Instrument administered by a sentencing team, comprising psychiatrist, clinical psychologist, social worker and criminologist, at lea st f or the most serious case. To test the validity and reliability of this new Instrument, the paper examines six very serious murder cases. Results of the analysis indicate that at least one case (where the person was executed in 2012) should have been reconsidered given a high score using the instrument (the higher score, the better the likelihood of rehabilitation and less risk of reoffending). Other murders in the sample with higher scores did receive a life sentence instead of death penalty. The paper argues that with limited data, the Instrument needs further empirical test. However, it is recommended that to provide for greater accuracy and consistency, courts employ such scientific instruments as a tool to assist sentencing/treatment decisions.