近年來，人工智慧（Artificial Intelligence, AI）與大型語言模型（Large Language Models, LLMs）如 ChatGPT、Gemini 等快速發展，也使語言與認知科學重新檢視一系列核心問題：人類如何理解語言？語言與概念如何在心智中表徵？語言理解是否仰賴預測？人類如何從經驗中習得語言結構與概念？人工智慧所展現的語言能力，如何幫助我們理解人類心智？
本課程以語言心理學（psycholinguistics）的核心問題為主軸，結合認知科學、人工智慧與計算模型的研究發展，探討語意表徵、詞彙與語句處理、語言學習，以及語言與知覺、社會互動與文化脈絡之間的關係。本課程不以程式設計或 AI 工程實作為主要內容，而是將不同類型與不同世代的計算模型（computational models）視為研究人類語言與認知的工具，探討模型如何被用來表徵、預測與模擬人類行為，如何產生新的心理語言學假設，以及模型結果能夠支持哪些理論推論、又有哪些限制。課程特別重視模型表現與人類心理機制之間的關係，引導學生思考：一個模型與人類表現相似時，我們究竟可以由此推論什麼？
課程採講授（lecture）與學生主導書報討論（student-led seminar）相結合的方式進行。學生每週課前預習一篇指定研究文獻，並且提問；各週由指定學生簡要介紹研究問題、方法與關鍵結果，並帶領全班針對文獻的理論意涵、模型與人類心智之間的推論，以及可能的替代解釋與後續研究進行討論。教師則於討論後進行理論統整與補充，並於課程尾聲提供下一週閱讀所需的核心概念與方法背景（conceptual primer），協助學生進入下一篇文獻，而不預先提供文獻的理論解讀。
透過反覆的閱讀、討論、批判與研究問題生成，學生將逐步學習如何判斷計算模型在心理語言學研究中所扮演的角色，區分模型表現、心理表徵與認知機制之間的不同層次，並評估模型結果對人類語言理論所能提供的證據與限制。課程最終將引導學生從每週文獻所產生的問題出發，發展自己的研究問題、理論假設與研究計畫。

Recent advances in artificial intelligence (AI) and large language models (LLMs), including systems such as ChatGPT and Gemini, have prompted researchers in language and cognitive science to revisit a range of fundamental questions: How is linguistic knowledge represented? To what extent does comprehension rely on prediction? How are linguistic structures and concepts acquired from experience? And what, if anything, can the success of contemporary AI systems tell us about human mind?
This course examines these questions through the lens of computational modeling. Rather than focusing on programming or AI engineering, we will study how computational models—from earlier cognitive and connectionist approaches to contemporary language and multimodal models—can be used to investigate human language and cognition. Psycholinguistic topics include semantic representation, lexical and sentence processing, language learning, multimodal cognition, and the social and cultural dimensions of language. A central theme throughout the course is the distinction between model performance and psychological explanation: when a model reproduces human-like behavior, what does that similarity allow us to conclude about human representations, learning processes, or cognitive mechanisms?
The course follows a hybrid lecture–seminar format. Students will read one research article each week and submit a brief reading preparation before class. Student discussion leaders will provide a concise introduction to the study and guide the class in evaluating its theoretical implications, the inferential links between model results and claims about the human mind, alternative explanations, and possible directions for future research. The instructor will conclude each discussion with theoretical synthesis and broader context, followed by a short conceptual primer introducing the background needed for the following week’s reading.
By the end of the course, students will be able to critically evaluate the role of computational models in psycholinguistic research, distinguish among behavioral, representational, and mechanistic levels of explanation, and assess what model results do—and do not—contribute to theories of human language. Students will also develop an original, model-inspired research proposal that translates insights from computational modeling into a testable question about human language and cognition.