Effects Of Emotion And Semantic Relatedness On Recognition Memory: Behavioral And Electrophysiological Evidence Part 2

2.1 | Method

2.1.1 | Participants

To determine the required sample size, we first extracted the behavioral effect size obtained by Pierce & Kensinger (2011; N = 32, Cohen's f = 0.42), who employed a similar experimental design to the one used in the current study.

There is a relationship between sample size and memory. This relationship may not be direct, but it is very important. Simply put, the larger the sample size, the more things we will remember. Therefore, we need to constantly collect new and meaningful information in our daily lives to improve our memory.

Scientific research shows that using different methods and channels to obtain information can stimulate our memory and improve our memory ability. Try to interact with people of different ages, carry out different leisure activities, and experience different cultures, environments, and people, so that we can make our knowledge more complete, broad, and deep.

Breaking through the scope of your knowledge and learning to memorize information quickly is also an important way to improve your memory. You can try to train your brain through some memory games, consolidate your memory through repeated memorization and review, and help you quickly master new knowledge. Improving memory does not require much cost or time, just maintain a positive attitude and keep trying.

In short, through continuous accumulation, training, learning, and trying, everyone can improve their memory, master more knowledge, and enrich their life experience. In this era of information explosion, it is very important to have a strong memory. We have the responsibility and obligation to improve our memory from all angles and become a better version of ourselves. It can be seen that we need to improve memory, and Cistanche deserticola can significantly improve memory, because Cistanche deserticola can also regulate the balance of neurotransmitters, such as increasing the levels of acetylcholine and growth factors. These substances are very important for memory and learning. In addition, Meat can also improve blood flow and promote oxygen delivery, which can ensure that the brain receives sufficient nutrients and energy, thereby improving brain vitality and endurance.

Click Know to improve short-term memory

Next, we extracted the associative memory ERP effects (for the early and late ERP components) from multiple studies that used an associative memory paradigm (Bader & Mecklinger,  2017; Kamp et al.,  2016; Li et al.,  2017; Li et al.,  2019; Rhodes & Donaldson, 2008; Tibon, Gronau, et al., 2014; Zhao et al., 2020; Zheng, Li, Xiao, Broster, Jiang, & Xi, 2015).

The effect sizes for the associative memory component reported in these studies were all medium-large, ranging from Cohen's f =  0.3 to 10.7 (N ranging between 17 and 46). 

Nevertheless, to avoid an overinflated estimation of effect size, we set f = 0.25 as a lower, more conservative value. Based on this effect size, we estimated that power>.8 (α =.01) would require at least 39 participants (actual power = .81) and therefore recruited 47 participants for the study.

Forty-seven healthy, right-handed native Chinese speakers (30 females; mean age 22.4 ±2 years) from Capital Normal University participated in the experiment and were paid ¥30/h. 

All participants had normal or corrected-to-normal vision and were prescreened for a history of neurological or psychiatric disorders, learning disorders, head injury, or psychotropic drug use. Informed consent, approved by the Capital Normal University Institutional Review Board, was collected from each participant. 

Data from five participants were discarded, including one participant with very poor task performance (associative Pr<0), and four participants with insufficient number of artifact-free ERP trials in one or more experimental conditions (N trial <16). Our final sample therefore included 42 participants (27 females; mean age 22.4 ±2 years).

Photoshop 8.0, and were presented at the center of the picture on a gray background (RGB: 150; see Figure 1 for examples).

An independent sample (N = 12) provided ratings of the pictures on the dimensions of valence, arousal, and familiarity, using a scale ranging from 1 (very negative/ calm/unfamiliar) to 9 (very positive/exciting/familiar). Pictures with familiarity ratings below 4 were removed from the pool. 

1272 pictures, including 646 negative pictures (with valence scores below 4) and 625 neutral pictures (with valence scores between 4 and 7), were chosen and combined to form negative/negative picture pairs or neutral/neutral picture pairs, resulting in 166 semantically-unrelated negative, 150 semantically unrelated neutral, 157 semantically-related negative, and 162 semantically-related neutral picture pairs.
Another independent sample (N =  10) provided ratings of the pairs on the dimension of relatedness. They were asked to judge how likely it is for the two objects to appear together (Tibon, Gronau, et al., 2014), by responding on a scale ranging from 1 (very unlikely) to 9 (very likely). 

Only pairs for which the pre-assigned relatedness status was verified (i.e., unrelated stimuli with a relatedness score<5, and related stimuli with a relatedness score≥5) by the majority of the raters (at least 6/10 raters) were included in the study. 

Based on the ratings, 600 pairs were selected, including 150 semantically related negative pairs, 150 semantically related neutral pairs, 150 semantically unrelated negative pairs, and 150 semantically unrelated neutral pairs. Semantically related pairs either belonged to the same category (e.g., "desk-sofa") or were functionally related (e.g., "rabbit-carrot"). 

The relatedness scores of related pairs [Mean (SD) = 6.61 (1.25)] were significantly higher than that of unrelated pairs [Mean (SD)  =  1.61 (.41); t (299)  =  64.20, p<.001]. 

Negative pictures were significantly more negative and arousing than neutral pictures [valence: Meaning (SD)  =  3.05 (0.66), Meanneu (SD)  =  5.06 (0.31), t (599)  =  −59.22, p<.001; arousal: Meaning (SD) =  5.43 (1.23), Meanneu (SD) =  2.41 (0.63), t (599) = 45.77, p<.001], but equal to neutral pictures on familiarity (p>.05).

We subsequently constructed 200 rearranged pairs by combining pictures belonging to different pairs, but keeping their type and location unchanged, such that there were 50 rearranged pairs for each type. For example, two related neutral pairs A-B (e.g., rabbit-carrot) and C-D (e.g., goat-cabbage) could be recombined to form another related neutral pair A-D (rabbit-cabbage). 

B and C would also be combined with other items (belonging to related neutral pairs) to form rearranged pairs. The same sample (N =  10) of participants in prior relatedness ratings were recruited again and provided ratings for relatedness. 

The results confirmed our initial assignment of pairs and showed that the relatedness scores of related pairs [Mean (SD) = 6.67 (1.23)] were significantly higher than that of unrelated pairs [Mean (SD) =  1.65 (.57); t (99) =  64.90, p <.001]. Importantly, there was no difference in relatedness between the rearranged pairs and the original pairs [Unrelated pairs: t (99)  =  1.07, p =.29; Related pairs: t (99) = 0.76, p =.35].

A total of 400 picture pairs were encoded at the study phase (100 related negative pairs, 100 related neutral pairs, 100 unrelated negative pairs, and 100 unrelated neutral pairs), with the remaining 200 pairs serving as new pairs during the test phase. At the test, 200 intact pairs (the same pairs shown in the study), 200 rearranged pairs (pictures belonging to different study pairs that were recombined together), and 200 new pairs were presented, with each condition containing 50 related negative, 50 related neutral, 50 unrelated negative, and 50 unrelated neutral pairs. Test pairs were counterbalanced across subjects, with every picture presented equally often as part of an intact, rearranged, or new pairing.

2.1.3 | Procedure

Participants were seated at a distance of 70 cm from a Dell monitor in an electrically shielded room. Picture pairs, with a visual area of 10°×5°, were displayed (using Presentation by Neurobehavioral Systems, Inc.) horizontally at the center of the monitor against a black background. A standard study-test paradigm was adopted, with the study phase followed by the test phase after a 10-minute delay. Four self-paced breaks were provided during the study phase and the test phase. Stimuli were presented pseudo-randomly to ensure that no more than three consecutive trials were from the same condition.

In the study, each trial began with a gray fixation cross for 1000–1500ms, followed by the presentation of a picture pair for 2000ms, during which the participants were asked to memorize the pairs and perform a valence judgment task, namely, to judge which one of the two objects is more negative (Figure 1b). They were asked to press the 'left arrow' key on the keyboard if they thought that the left object was more negative, to press the 'right arrow' key if they thought that the right one was more negative, and to press the 'down arrow' key if they thought that the two objects had similar valence. 

Once the study phase was completed, a 10-minute break was provided. During this period, participants performed a distractor task of 3- 3-digit backward counting for 5 min and then rested for five additional minutes.

At the test, each trial began with a jittered fixation cross presented for 1000–1500ms, followed by the presentation of a picture pair for 2000ms. Participants were asked to indicate whether the pair is "intact", "rearranged", or "new" as accurately and as quickly as possible. Responses were provided via keyboard keys, counterbalanced across participants. 

Half of the participants made responses of "intact" and "rearranged" by pressing the key "F" and "G" with the left hand, and "new" by pressing the key "J" with the right hand. The other half of the participants responded "intact" and "rearranged" by pressing the key "H" and "J" with the right hand, and "new" by pressing the key "F" with the left hand.

A study practice block of 12 trials was provided at the beginning of the experiment, before the study phase. An additional test practice block of 18 trials was provided before the test phase. During these practice sessions, the experimenter ascertained that the participants understood the task.

2.1.4 | EEG recording and preprocessing

EEG was recorded using a 64-channel Neuroscan system and the electrode locations adhered to the extended international 10–20 system. The sampling rate was 500Hz with a 0.05–100Hz bandpass filter. Electrooculogram (EOG) was recorded using two electrodes placed outside the outer canthi of each eye and one infraorbital to the left eye. 

The left mastoid was used as the reference site online, and EEG signals were re-referenced offline to the average of the left and right mastoid recordings. Impedance was kept below 5 kΩ. EEG/EOG signals were filtered with a bandpass of 0.05–40Hz. EEG data from the test phase were segmented into 1100 ms epochs, corrected to the 100 ms pre-stimulus baseline. Epochs with a voltage exceeding ±75 μV were excluded. 

Independent component analysis (ICA) conducted with the runica algorithm available through the EEGLAB toolbox for MATLAB (v.2019.0, Delorme & Makeig, 2004), was used to isolate and remove EOG blink artifacts. A minimum of 16 trials for each condition was required to ensure an acceptable signal-to-noise ratio, and four participants were excluded for failing to meet the minimal number of trials.

Mean numbers of related analyzed trials were 39 (intact), 26 (rearranged), and 38 (new) for negative pairs, and 37 (intact), 26 (rearranged), and 43 (new) for neutral pairs. The mean numbers of unrelated analyzed trials were 25 (intact), 30 (rearranged), and 37 (new) for negative pairs, and 27 (intact), 30 (rearranged), and 41 (new) for neutral pairs.

2.1.5 | Statistical analyses: General approach

Data were extracted for correct trials only (e.g., Donaldson & Rugg,  1998; Paller et al.,  2003). Repeated measures analyses of variance (ANOVAs) were conducted for inferential statistics, with Greenhouse–Geisser correction for non-sphericity when required. Follow-up analyses were performed using repeated measures ANOVAs or t-tests as appropriate. 

To control for Type I error rates, p-values were corrected for false discovery rate (FDR) with the Benjamini–Hochberg procedure (Benjamini & Hochberg,  1995) at p <.05. Because the current study focuses on mnemonic effects, only main effects and interactions that included the factor of response type are reported.

Behavioral analyses

The behavioral measure of interest was associative Pr: a discrimination measure of old/new effects for associative memory, defined by subtracting false alarm rates for rearranged pairs from hit rates for intact pairs (Jäger et al., 2006; Snodgrass & Corwin, 1988). This measure was used to dissociate potential response bias (e.g., for related pairs; Ahmad & Hockley, 2014; Liu & Guo, 2019; Tibon, Gronau, et al., 2014; see Supporting Information 1 for an ancillary analysis of response bias) from a true memory advantage. Pr scores were analyzed using a repeated measure ANOVA with relatedness (related or unrelated) and valence (negative or neutral) as repeated factors and with the Pr score as the dependent measure.

Given our interest in associative memory, and disentangling true memory effects from response bias, we further analyzed accuracy rates (% correct) for intact pairs and rearranged pairs using a repeated ANOVA with relatedness (related or unrelated) and valence (negative or neutral) as repeated factors, and accuracy rate as the dependent measure. A full 3-way ANOVA which includes all factors (relatedness, valence, and response type) and levels (intact, rearranged, new) within the same model, is included in Supporting Information 2.

ERP analyses

Both intact and rearranged pairs are comprised of studied items. However, while intact pairs further contain studied associative information, rearranged pairs contain novel associative information which was not presented in the study. Therefore, the intact/rearranged effects, i.e., differences between ERPs associated with correct "intact" judgments vs. correct "rearranged" judgments, are indicative of associative recognition. (e.g., Li et al., 2017; Rhodes & Donaldson,  2008; Zheng, Li, Xiao, Broster, Jiang, & Xi,  2015). 

Accordingly, we focused on the comparison between intact and rearranged pairs to index associative memory. For completion, we also include the comparison between rearranged and new pairs as an index of item memory in Supporting Information 3.

For the frontal and parietal memory effects, time segments and regions of interest were defined based on previous ERP studies (Bader et al., 2010; Han et al., 2018; Li et al., 2017, 2019; Rugg & Curran,  2007; Speer & Curran,  2007; Wolk et al., 2006; Zheng et al., 2016). Accordingly, two-time windows, 300–550ms and 550–800ms, were used to capture the frontal and parietal memory effects, respectively. Mean amplitudes for statistical analyses in these windows were obtained from frontal (collapsed across F3, Fz, and F4), central (collapsed across C3, Cz, and C4), and parietal (collapsed across P3, Pz, and P4) scalp locations (Han et al., 2018; Hou et al., 2013; Molinaro et al., 2011).

Repeated measures ANOVA was conducted separately for each time window and included four within-subjects factors: relatedness (related or unrelated), valence (negative or neutral), response type (intact or rearranged), and location (frontal, central, or parietal).

2.2 | Results

2.2.1 | Behavioral results

Means and SDs for the various behavioral measures of Experiment 1 are shown in Table 1. The ANOVA for associative Pr (relatedness × valence) revealed a significant main effect of valence, F (1, 41) =  24.28, p <.001, 휂2 p =0.37 (greater Pr scores for neutral vs. negative pairs), and of relatedness, F (1, 41) = 120.01, p <.001, 휂2 p =0.75 (greater Pr scores for related vs. unrelated pairs). 

The analysis further revealed a significant 2-way interaction between relatedness and valence, F (1, 41) = 6.61, p =.014, 휂2 p =0.14, with lower associative Pr for negative pairs (vs. neutral) in the unrelated condition, t (41) = 5.38, p <.001, d = 0.83, but not in the related condition, t (41) = 1.65, p =.107, d = 0.25.

The analysis of accuracy rates for intact pairs ("hits") revealed a main effect of relatedness, F (1, 41) = 361.91, p <.001, 휂2 p =0.90, and a 2-way interaction between the two factors, F (1, 41) =  17.35, p <.001, 휂2 p =0.30, resulting from lower accuracy rates for negative pairs (vs. neutral) in the unrelated condition, t (41) =  2.78, p =.008, d = 0.43, but greater accuracy rates for negative pairs (vs. neutral) in the related condition, t (41) = 2.31, p =.026, d =  0.36. The analysis of accuracy rates for rearranged pairs ("correct rejections") only revealed a main effect of relatedness, F (1, 41) = 22.70, p <.001, 휂2 p =0.36, with greater accuracy rates for unrelated vs. related pairs.

Taken together, the behavioral results depict the predicted emotional associative interference effect, indicated by reduced Pr scores and accuracy rates for negative vs. neutral pairs. Furthermore, this effect was attenuated by semantic relatedness, with greater emotional interference observed for unrelated vs. related pairs.

2.2.2 | ERP results

Waveforms and topographical distribution of the associative memory effect in the various experimental conditions are shown in Figure 2. In the early time window (300–550 ms), the ANOVA for the associative memory effect revealed a 2-way interaction between valence and response type, F (1, 41) = 9.29, p =.004, 휂2 p =0.19. 

Decomposition of the interaction using paired t-tests at each level of valence, revealed a significant associative memory effect (more positive-going waveforms for intact vs. rearranged) for negative pairs, t (41) = 3.13, p =.003, d =  0.48, but not for neutral pairs (p =.29). Thus, in the early time-window, the associative memory effect emerged for negative pairs, regardless their relatedness, and had widespread distribution. 

In the late time window (550–800 ms), the ANOVA only revealed a main effect of response type, F (1, 41) = 38.84, p <.001, 휂2 p =0.49, suggesting that the late associative memory effect was similarly observed in all conditions and all locations. Exploratory analysis of a later associative memory effect (800–1000 ms), which resembled the pattern observed in the 550–800 ms time window, is included in Supporting Information 4.

For more information:1950477648nn@gmail.com