关键词: 词汇学习, 快速映射, 学习理论

Abstract: Fast mapping (FM) learning refers to a learning strategy through which learners quickly establish a connection between a novel word and its referent by excluding known words. The concept of “fast mapping” originates from children’s learning of native language vocabulary, in which 3- to 4-year-old children are able to establish the association between a new word (chromium) and an object (olive green) by excluding a known word (red), thus realizing the rapid acquisition of the meaning of a new word. Research on FM learning suggests that new words can rapidly form neural representations in the brain, challenging traditional learning theories and attracting considerable research interest.
The current paper reviews recent advances in FM learning from three aspects. First, FM learning demonstrates marked advantages over explicit encoding learning. From infancy, children can rapidly acquire new words through this mechanism; and in adults, FM learning similarly supports the rapid formation and long-term maintenance of word-object associations, with memory exhibiting slower decay and greater stability over time. Second, the underlying cognitive mechanisms of FM learning are examined in detail with reference to its key moderating factors. Specifically, from the perspective of exclusion-based reasoning during the learning process, it is demonstrated that disambiguation contributes to vocabulary acquisition, and that the mechanisms of active decision-making and response selection further facilitate the learning outcomes of FM learning. From the perspective of learners’ prior knowledge, the intrinsic connection between new and existing knowledge is shown to promote the rapid formation of new word memory during FM learning. Finally, a direct neural comparison between FM learning and explicit encoding learning reveals that FM learning relies on a cortical network anchored in the anterior temporal lobe (ATL), with little or even no hippocampal involvement—a pattern that is fundamentally distinct from the hippocampal-prefrontal system that underlies explicit encoding learning.
Since it was first proposed, FM learning has been met with questioning and challenges from researchers. The controversies surrounding FM learning can be broadly summarized into three major aspects. The first concerns whether FM learning is unique. At both the behavioral and neural levels, there has been conflicting evidence regarding the learning efficacy and “independence from the hippocampus” of FM learning. However, the uniqueness debate does not negate the encoding mode of FM learning itself, but rather lies in whether the engagement of the “cortical shortcut” may be constrained by the functional state of the hippocampus. The second concerns whether FM learning differs across different types of vocabulary learning. Vocabulary learning can be divided into two aspects: lexical configuration and lexical engagement. FM learning and explicit encoding learning have different applicability to these two aspects, which may be the source of the inconsistencies in previous findings. The third concerns whether FM learning challenges traditional learning theories. The rapid cortical integration mechanism revealed by FM learning poses a challenge to the systems consolidation view advocated by complementary learning systems and standard consolidation theory, whereas memory systems reorganization theory provides a more explanatory framework for these findings.
Based on this, future research on FM learning should be further advanced in the following aspects. At the cognitive-behavioral level, the underlying mechanisms by which FM learning facilitates cortical memory need to be thoroughly examined; at the neurophysiological level, techniques from cognitive neuroscience should be integrated to uncover the neural mechanisms of FM learning—for example, distinguishing different cognitive processes during FM learning and revealing dynamic interactions between brain regions through neural oscillatory dynamics, and establishing the causal roles of key brain regions in FM learning using neuromodulation techniques such as tDCS and TMS. Critically, the debates on FM learning have collectively driven learning theory from a linear consolidation view that asks “when information transfers from the hippocampus to the cortex” toward a dynamic reorganization view that asks “under what conditions the cortex can directly participate in the construction of novel word representations.” Future research needs to further clarify the cognitive and neural mechanisms of FM learning within this dynamic perspective, thereby providing empirical foundations for the development of learning theory.

Key words: vocabulary learning, fast mapping, learning theory