Introduction

Breast cancer is the most commonly diagnosed cancer among women worldwide, accounting for one in every four diagnoses in women [1]. Although cancer treatments have substantially improved survival rates [2], growing evidence highlights treatment-related cognitive side effects, particularly with chemotherapy [3]. Colloquially known as “chemobrain”, chemotherapy-related difficulties in memory, executive function, and processing speed [4] commonly emerge during or after treatment and may persist for years [5], 6]. These cognitive changes can hinder interpersonal, occupational, and everyday functioning, and ultimately reduce quality of life [7].

Memory impairment, including episodic memory, is frequently reported by breast cancer survivors (BCS) [8], 9]. Episodic memory involves the vivid recollection of personal events, including temporal, contextual, and perceptual details (e.g., recalling the last Olympic gold medal hockey game) [10]. While rich in detail, episodic memories can co-exist with gist-like versions that convey general event content but lack specific, vivid features [11], 12]. Both fine-grained episodic details and more coarsely-grained gist-like details depend on the hippocampus, engaging posterior and anterior regions respectively [13], [14], [15]. Over time, as a memory becomes less episodically-detailed and more schematized (generalized representation of a class of event; e.g., memory of a typical hockey game) [16], 17], its retrieval may become increasingly supported by the medial prefrontal cortex (mPFC) and less dependent upon detailed hippocampal representations [11], [18], [19], [20]. In contrast, semantic memory refers to general knowledge and facts (e.g., knowing each team has six players on the ice), and is not linked to a specific event [10]. Due to episodic memory’s reliance on the hippocampus for encoding and detailed retrieval, episodic memory is especially vulnerable to conditions that affect hippocampal integrity [18]. Consequently, individuals with medial temporal lobe damage often show impaired recollection of episodically-specific event details while relying on more generalized, mPFC-dependent schematic representations and semantic knowledge to support memory retrieval [12].

Numerous classes of chemotherapeutic drugs are neurotoxic to the hippocampus, causing structural (e.g., volume reduction, deformation, neurogenesis) [8], [21], [22], [23], [24] and functional (e.g., altered activity and connectivity) [23], 25], 26] hippocampal disruptions, offering potential mechanisms underlying chemotherapy-related episodic memory deficits [27]. Along the hippocampal long-axis, reduced posterior hippocampal volume predicted impoverished episodic autobiographical memory in breast cancer survivors [21], suggesting that episodically-detailed narrative memory is particularly vulnerable to impairment post-treatment, while more schematic representations may be less affected.

Conventional memory assessments used to assess episodic memory, including list learning and verbal paired associate tasks, though reliable, often lack ecological validity and may not reflect real-world cognitive functioning [28]. For instance, list-learning tasks assess immediate recall but lack everyday demands such as remembering conversations or personal events. BCS often show poorer performance on early list learning trials, suggesting initial encoding challenges that improve with repetition and familiarity [29], [30], [31]. However, some show delayed recall impairments despite intact immediate recall and recognition, suggesting deficits in retrieval rather than encoding deficits [6], 31]. Other studies report no differences between BCS and controls [32], 33], highlighting the complexity of cognitive impairments and the need for assessments that better characterize memory processing differences with a higher degree of sensitivity.

Unlike the arbitrary word pairs used in verbal paired associates tasks, narrative memory assessments engage more complex mnemonic processes including schematic and semantic representations, and require the construction and maintenance of complex temporal and contextual relational structures within an event [34]. Consequently, narrative memory tasks offer greater ecological validity and are more sensitive to detecting subtle memory disruptions across multiple mnemonic domains. Naturalistic memory tasks like narrative recall, better mimic real-life memory experiences and can evaluate precise verbatim recollection, gist-based recall, and memory distortions during recollection of complex events [35]. This is meaningful as verbatim memory typically declines with age, while gist memory remains more stable and less affected by hippocampal damage [35], 36]. Supporting this, a study using the Wechsler Memory Scale – Logical Memory (WMS – LM) story recall task found that BCS who were 5–15 years post-treatment initially recalled fewer narrative details than non-cancer controls (NC), although this difference resolved eight months following baseline [29]. By using unit-based scoring of narrative recollection, including verbatim, gist, and distortion sub-types, we can potentially detect more subtle memory impairments that may not be captured by dichotomous scoring (i.e., correct or incorrect) [37] used in standardized memory assessments like verbal paired associates and the original scoring for the WMS-LM, leading to an underestimate of cognitive impairment.

Compounding this challenge, BCS often experience subjective cognitive decline post-chemotherapy [9], even in cases which fail to show impairments using objective measures [38]. A potential variable contributing to this discrepancy is perceived cognitive workload, or the mental effort required to complete a task [39], which is influenced by cognitive reserve, task complexity, and neural efficiency [40]. In addition to the hippocampal neurotoxicity discussed above, chemotherapy treatment leads to broader structural and functional changes throughout the brain (e.g., prefrontal hyperactivation, decreased grey and white matter integrity, and alterations in the default mode network) [41], [42], [43], which may contribute to inefficient network processing during cognitive tasks. The mismatch between objective and subjective complaints by cancer survivors may reflect the need to engage energy-intensive neural compensatory mechanisms in order to effortfully perform at control levels [44], 45].

This study examined whether an ecologically valid measure of narrative event memory could identify subtle memory deficits that may not be captured by a standard episodic memory assessment. We hypothesized that BCS would recall fewer episodically rich (i.e., verbatim) details during recall of the Taler Stories relative to NC. Additionally, we expected BCS to exhibit more difficulty recalling word pairs during the first trial of the Verbal Paired Associates (VPA) task, with improvements across trials. Memory performance was predicted to be associated with higher cognitive effort reported by BCS.