The First Word Recalled Measure - A Potential Addition to Clinical Exams

Abstract

Characterizing episodic memory abilities is highly important in the diagnosis of Alzheimer's disease (AD) and mild cognitive impairment (MCI), and usually includes wordlist learning and recall tasks. Clinical evaluations typically focus on the number of words recalled, ignoring additional information, like serial position. Here, we tested the potential value of two serial positioning measures for clinical diagnosis - how retrieval is initiated, as measured by the first word recalled, and how it proceeds - using data from patients with AD and MCI that completed a wordlist learning and recall task. Our results show that during the early stages of learning, patients with AD are less prone to retrieve the first word from the wordlist, manifested as lower primacy effect in the first word recalled, compared with MCI patients. The first word recalled measure adds to the differentiation between the groups over and above the total number of words learned. Thus, the first word recalled during word list learning and recall tasks may be used as a simple complementary measure to distinguish between MCI and AD during standard neuropsychological evaluations.

Keywords: Alzheimer disease; differential diagnosis; memory; mild cognitive impairment; neuropsychological tests.

Figure 1

Serial position curve, representing retrieval chances as a function of serial position in the wordlist during encoding. Graphs (A,B) show overall probabilities, (C,D) show probabilities for first word recalled. In immediate recall (CERAD 1, 2, 3, gray-scale graph lines), both groups showed primacy and recency effects in both overall probability (A,B) and first word recalled (C,D). In delayed recall (CERAD-DR, red), none of the groups showed recency effect (C,D), as expected. In contrast, there is a seemingly considerable reduction in the dementia group in primacy effect during delayed recall, as compared to other learning trials (D), which is less evident in the MCI group (C).

Figure 2

The distribution and proportion of first word recalled. (A) Presents the percent of “First” (F), “Middle-list” (M), and “Recent” (R) responses retrieved as the first word, relative to the expected percent based on uniform distribution (see the “Exp” bar on the left). The percent is calculated after discarding “None” responses. Sections denoted with thick borders indicate significant difference from the expected distribution. (B) Presents the proportion of “None” responses (checkered) relative to recalls (first recall, black) in each group and recall trial [CERAD 1-3 and delayed recall (CERAD-DR)].

Figure 3

The mean proportion of each possible transition between items as a function of the lag between items in the encoded wordlist. Results indicate that both groups maintained the tendency to retrieve items from nearby serial position of the just-recalled word, as well as the typical asymmetry reflected as greater proportion of forward transitions as compared to backward transitions. Dashed red line present data from a sub group of patients that retrieved at least two items during delayed recall (N = 17 patients with AD, 26 patients with MCI).