Saccades and handedness interact to affect scene memory

Abstract

Repetitive saccades benefit memory when executed before retrieval, with greatest effects for episodic memory in consistent-handers. Questions remain including how saccades affect scene memory, an important visual component of episodic memory. The present study tested how repetitive saccades affect working and recognition memory for novel scenes. Handedness direction (left-right) and degree (strong/consistent vs. mixed/inconsistent) was measured by raw and absolute laterality quotients respectively from an 8-question handedness inventory completed by 111 adults. Each then performed either 30 s of repetitive horizontal saccades or fixation before or after tasks of scene working memory and scene recognition. Regression with criterion variables of overall percent correct accuracy and d-prime sensitivity showed that when saccades were made before working memory, there was better overall accuracy as a function of increased direction but not degree of handedness. Subjects who made saccades before working memory also performed worse during subsequent recognition memory, while subjects who fixated or made saccades after the working memory task performed better. Saccades made before recognition resulted in recognition accuracy that was better (Cohen's d = 0.3729), but not significantly different from fixation before recognition. The results demonstrate saccades and handedness interact to affect scene memory with larger effects on encoding than recognition. Saccades before scene encoding in working memory are detrimental to short- and long-term memory, especially for those who are not consistently right-handed, while saccade execution before scene recognition does not appear to benefit recognition accuracy. The findings are discussed with respect to theories of interhemispheric interaction and control of visuospatial attention.

Keywords: Encoding; Eye movements; Laterality; Recognition; Regression analyses; Working memory.

Figure 1. Example scene working and recognition memory trials.

Each working memory trial (A) consisted of an encoding phase of five scene stimuli, each presented for 2 s, followed by a 6 s delay period of crosshair fixation and a 2 s presentation of either a positive probe (scene from the encoding set) or a negative probe (new scene). On each trial, there was a 50% chance of a positive probe appearing. Each recognition memory trial (B) consisted of alternating presentation of a scene stimuli for 2 s followed by a black screen for 2 s. Old and new scenes were randomly intermixed. Subjects were instructed to press a green button if they had seen a stimulus in any of the previous working memory trials and a red button if they had never seen the stimulus before. In this example, image numbers from the set of scene stimuli are listed rather than the actual scene images to illustrate how scenes during encoding could be presented as a positive probe (A) or as old stimuli in the recognition task (B).

Figure 2. The Saccade and Fixation Eye Tasks.

The saccade task (A) required subjects to alternate for 30 s looking toward the left and right as a white disc moved back and forth across (B) show saccadic movements with large periodic deviations in the x position (horizontal) trace with a stable y position (vertical) trace. The fixation task (C) required subjects to maintain fixation on a center disc as it changed color. Example eye-tracking traces (D) show fixation, with minimal deviations in the x (horizontal) and y position (vertical) traces.

Figure 3. Scene working memory performance as a function of laterality and eye task timing.

There was a significant relationship in working memory percent correct performance as a function of direction of laterality when saccades were made before the working memory task (non-zero slope of 0.1029 ± 0.04365 SE, F_(1,30) = 5.56, p = 0.0251, red line and red triangles, A). Overall mean performance among conditions differed (non-zero intercepts, F_(3,106) = 6.164, p = 0.0007) with lowest working memory percent correct performance obtained when saccades were made before the working memory task (77.79% ± 3.84, 95% CI [69.95–85.64%]) and best performance when saccades were made after the working memory task (93.28% ± 1.844 SE, 95% CI [89.45–97.12]). (B) shows the relationship of the sensitivity measured d-prime with direction of laterality, while (C) and (D) show relationships between percent correct accuracy and d-prime respectively with degree of laterality as expressed by the absolute value of laterality quotients.

Figure 4. Scene Recognition Memory Performance as a Function of Laterality and Eye Task Timing.

There was a significant relationship in recognition percent correct performance as a function of direction of laterality during the condition of fixation before the recognition memory task (non-zero slope = 0.08738 ± 0.03277 SE, F_(1,21) = 7.111, p = 0.0144, grey line and grey boxes (A). Overall mean performance among the conditions differed (F_(3,106) = 4.721, p = 0.0039) with best recognition memory performance during the condition of fixation before the working memory task (70.08% ± 3.899 SE, 95% CI [62.12–78.03], blue line and blue circles, A) and lowest recognition memory performance during the condition when saccades were made before the working memory task (60.67% ± 4.06, 95% CI [52.38–68.96], red line and red triangles, A). (B) shows the relationship of the sensitivity measured d-prime with direction of laterality, while (C) and (D) show relationships between percent correct accuracy and d-prime respectively with degree of laterality as expressed by the absolute value of laterality quotients.