Common use of Stimuli Clause in Contracts

Stimuli. Norms for name agreement, as well as for other attributes for the pictures and picture names were obtained from ▇▇▇▇▇▇▇▇, ▇▇▇▇▇▇▇▇ and ▇▇▇▇▇‘s database (1997). Images were selected to be of high or low name agreement whilst matched on other attributes (see Table 1). Stimuli were selected to yield 50 low name agreement items, 50 high name agreement items, eight practice trials and 30 fillers (see Appendix for a list of words used in the experiment). Corresponding pictures were chosen, 114 from original ▇▇▇▇▇▇▇▇▇ and Vanderwart‘s pictures (1980), eight from ▇▇▇▇▇▇▇▇▇ and Vanderwart pictures redrawn by ▇▇▇▇▇▇▇▇ et al. (1997) and 16 from additional drawn pictures by ▇▇▇▇▇▇▇▇ et al. (1997). ▇▇▇▇▇▇▇▇▇ and Vanderwart‘s original pictures (1980) were digitally scanned into TIF format documents and processed into clear line-drawings with high contrast. ▇▇▇▇▇▇▇▇ et al.‘s additional pictures were downloaded from ▇▇▇▇▇▇▇▇‘s online source, ▇▇▇▇://▇▇▇.▇▇.▇▇.▇▇/uwcc/psych/▇▇▇▇▇▇▇▇/ (no longer available at this address). All pictures were saved as bitmaps (see Figure 1 for examples). On-screen size of all pictures was smaller than 3cm horizontally and 3.5cm vertically. The DPI (dots per inch) value for all pictures was 72 pixels/inch. INSERT TABLE 1 ABOUT HERE From the original database (▇▇▇▇▇▇▇▇ et al., 1997), the 50 low name agreement (‗LNA‘) items had name agreement percentages between 50% and 87% (Mean=76.0%, SD= 10.2%). The 50 high name agreement (‗HNA‘) items all had 100% name agreement. Independent t tests revealed that they differed significantly in name agreement, but not in other attributes (see Table 1). We additionally checked whether the three sets of images (LNA, HNA, fillers) differed from each other in terms of low level picture attributes. Using data from a previous study by Laws and ▇▇▇▇ (2002) we established two intrinsic measures for each image: (1) the proportion of black pixels, and (2) the internal complexity of each image—the latter measure Laws and ▇▇▇▇ (2002) found to vary systematically for different categories of the ▇▇▇▇▇▇▇▇▇ pictures. Data were available for the majority of our pictures (LNA 42/50, HNA 37/50, fillers 24/30). Using MANOVA to compare the two image attributes simultaneously across the three picture groups, we found no difference between the image sets, ▇▇▇▇‘s Lambda F(4, 198) = 1.06, p = .378. We concluded that the low level attributes of the image sets did not differ systematically. Eight practice trials were devised, four with low name agreement images and four with high name agreement images. A further 30 filler (catch) trials contained images in which the names and pictures presented did not match. Stimuli were presented using the E-prime package (version 1.2, Psychology Software Tools, Inc., Pittsburgh, PA) on a 17‖ CRT screen. The visual angles were moderated to be less than 7o horizontally and 8o vertically (▇▇▇▇▇▇▇ & ▇▇▇▇▇▇▇▇▇, 1994). Participants sat in front of the monitor and read the instructions, then pressed the space bar to initiate each trial. A practice session with eight trials was followed by an experimental session. Each trial commenced with the presentation of a small black fixation cross in the center of the monitor (duration 1500ms). Subsequently, a picture was shown on the screen for 1000 ms and participants were asked to name the picture covertly (i.e. silently, to themselves) as soon as it was presented. After picture offset, the most common name for the picture was presented visually, with the question ―Same name?‖ underneath, which, as the participants had previously been informed, prompted them to decide whether the name on the screen exactly matched the word in their heads. Participants responded ―Yes‖ with their left hand or ―No‖ with their right hand by keypress. Participants were asked not to blink their eyes or to move any part of their body during the time the pictures were on the screen. All trials (50 HNA items, 50 LNA items and 30 fillers) were presented once in random order. For high and low name agreement items, it was highly likely that the expected name shown on the screen would be the same as that the participants had named covertly, so most answers would be ―Yes‖. For this reason, filler (catch) trials were introduced, so that participants had to think before responding. Participants pressed the space bar to start each trial and they could take a break whenever necessary by not pressing the space bar. The whole recording session took approximately 10 minutes. Electrophysiological (EEG) signals were collected from the scalp with an Electrical Geodesics GSN 200 sensor net system with 128 channels (Electrical Geodesics, Inc., Eugene, OR), amplified by the EGI NetAmps 200 high impedance amplifier with a bandpass of 0.1 – 100 Hz, and digitized at a sample rate of 250 Hz. The threshold for impedance was set at 50 kΩ and all sites were recorded with a vertex reference. Electrophysiological signals were filtered with a 40 Hz low-pass filter. In order to correct the polar average reference effect (PARE), a PARE-corrected reference was used (Junghofer, Elbert, Tucker, & ▇▇▇▇▇, 1999), which was computed from the average of the entire surface of the scalp. Subsequently, individual trials were labeled as bad because of eye blinks or movements (EOG was recorded from 6 electrodes: 8, 26, 125, 126, 127, 128. An eye-blink or eye-movement is identified when amplitude is over 70μV in any of the six electrodes.) or because of more than 10 bad channels (defined as having an average amplitude over 200μV or surpassing a differential threshold of over 100μV, compared to a 10-sample running average). In the final data analysis, 95.4% of trials were good in the HNA group and 95.2% in the LNA group (good trials rates ranged from 90% to 100%). Bad channels were replaced by the interpolation of good channels in proximity to the bad ones according to the spherical spline algorithm (Srinivasan, ▇▇▇▇▇, ▇▇▇▇▇▇, ▇▇▇▇▇▇▇▇▇▇▇, & ▇▇▇▇▇▇▇, ▇▇▇▇). After bad channel replacement, all segments for each condition of each participant were averaged individually. Finally, ERPs were baseline corrected using a 200 ms pre-stimulus interval. The epoch length was 1200ms. The purpose of the post test, which took place immediately after ERP recording, was to classify images according to the source of name disagreement. Accordingly, only low name agreement pictures were used. Participants were instructed to press a key to indicate the relationship between a picture and a written name. In each trial, a low name agreement picture was presented with its name (according to ▇▇▇▇▇▇▇▇ et al., 1997), and two choices were given: (1) ―This is ONE OF the names of the object‖ (i.e., the disagreement source is alternative names for the object depicted (AN subgroup); or

Stimuli. Norms for name agreement, as well as for other attributes for the pictures and picture names were obtained from ▇▇▇▇▇▇▇▇, ▇▇▇▇▇▇▇▇ and ▇▇▇▇▇‘s database (1997). Images were selected to be of high or low name agreement whilst matched on other attributes (see Table 1). Stimuli were selected to yield 50 low name agreement items, 50 high name agreement items, eight practice trials and 30 fillers (see Appendix for a list of words used in the experiment). Corresponding pictures were chosen, 114 from original ▇▇▇▇▇▇▇▇▇ and Vanderwart‘s ▇▇▇▇▇▇▇▇▇▇‘s pictures (1980), eight from ▇▇▇▇▇▇▇▇▇ and Vanderwart pictures redrawn by ▇▇▇▇▇▇▇▇ et al. (1997) and 16 from additional drawn pictures by ▇▇▇▇▇▇▇▇ et al. (1997). ▇▇▇▇▇▇▇▇▇ and Vanderwart‘s ▇▇▇▇▇▇▇▇▇▇‘s original pictures (1980) were digitally scanned into TIF format documents and processed into clear line-drawings with high contrast. ▇▇▇▇▇▇▇▇ et al.‘s additional pictures were downloaded from ▇▇▇▇▇▇▇▇‘s online source, ▇▇▇▇://▇▇▇.▇▇.▇▇.▇▇/uwcc/psych/▇▇▇▇▇▇▇▇/ (no longer available at this address). All pictures were saved as bitmaps (see Figure 1 for examples). On-screen size of all pictures was smaller than 3cm horizontally and 3.5cm vertically. The DPI (dots per inch) value for all pictures was 72 pixels/inch. INSERT TABLE 1 ABOUT HERE From the original database (▇▇▇▇▇▇▇▇ et al., 1997), the 50 low name agreement (‗LNA‘) items had name agreement percentages between 50% and 87% (Mean=76.0%, SD= 10.2%). The 50 high name agreement (‗HNA‘) items all had 100% name agreement. Independent t tests revealed that they differed significantly in name agreement, but not in other attributes (see Table 1). We additionally checked whether the three sets of images (LNA, HNA, fillers) differed from each other in terms of low level picture attributes. Using data from a previous study by Laws ▇▇▇▇ and ▇▇▇▇ (2002) we established two intrinsic measures for each image: (1) the proportion of black pixels, and (2) the internal complexity of each image—the latter measure Laws and ▇▇▇▇ (2002) found to vary systematically for different categories of the ▇▇▇▇▇▇▇▇▇ pictures. Data were available for the majority of our pictures (LNA 42/50, HNA 37/50, fillers 24/30). Using MANOVA to compare the two image attributes simultaneously across the three picture groups, we found no difference between the image sets, ▇▇▇▇‘s Lambda F(4, 198) = 1.06, p = .378. We concluded that the low level attributes of the image sets did not differ systematically. Eight practice trials were devised, four with low name agreement images and four with high name agreement images. A further 30 filler (catch) trials contained images in which the names and pictures presented did not match. Stimuli were presented using the E-prime package (version 1.2, Psychology Software Tools, Inc., Pittsburgh, PA) on a 17‖ CRT screen. The visual angles were moderated to be less than 7o horizontally and 8o vertically (▇▇▇▇▇▇▇ & ▇▇▇▇▇▇▇▇▇, 1994). Participants sat in front of the monitor and read the instructions, then pressed the space bar to initiate each trial. A practice session with eight trials was followed by an experimental session. Each trial commenced with the presentation of a small black fixation cross in the center of the monitor (duration 1500ms). Subsequently, a picture was shown on the screen for 1000 ms and participants were asked to name the picture covertly (i.e. silently, to themselves) as soon as it was presented. After picture offset, the most common name for the picture was presented visually, with the question ―Same name?‖ underneath, which, as the participants had previously been informed, prompted them to decide whether the name on the screen exactly matched the word in their heads. Participants responded ―Yes‖ with their left hand or ―No‖ with their right hand by keypress. Participants were asked not to blink their eyes or to move any part of their body during the time the pictures were on the screen. All trials (50 HNA items, 50 LNA items and 30 fillers) were presented once in random order. For high and low name agreement items, it was highly likely that the expected name shown on the screen would be the same as that the participants had named covertly, so most answers would be ―Yes‖. For this reason, filler (catch) trials were introduced, so that participants had to think before responding. Participants pressed the space bar to start each trial and they could take a break whenever necessary by not pressing the space bar. The whole recording session took approximately 10 minutes. Electrophysiological (EEG) signals were collected from the scalp with an Electrical Geodesics GSN 200 sensor net system with 128 channels (Electrical Geodesics, Inc., Eugene, OR), amplified by the EGI NetAmps 200 high impedance amplifier with a bandpass of 0.1 – 100 Hz, and digitized at a sample rate of 250 Hz. The threshold for impedance was set at 50 kΩ and all sites were recorded with a vertex reference. Electrophysiological signals were filtered with a 40 Hz low-pass filter. In order to correct the polar average reference effect (PARE), a PARE-corrected reference was used (Junghofer▇▇▇▇▇▇▇▇▇, Elbert▇▇▇▇▇▇, Tucker▇▇▇▇▇▇, & ▇▇▇▇▇, 1999), which was computed from the average of the entire surface of the scalp. Subsequently, individual trials were labeled as bad because of eye blinks or movements (EOG was recorded from 6 electrodes: 8, 26, 125, 126, 127, 128. An eye-blink or eye-movement is identified when amplitude is over 70μV in any of the six electrodes.) or because of more than 10 bad channels (defined as having an average amplitude over 200μV or surpassing a differential threshold of over 100μV, compared to a 10-sample running average). In the final data analysis, 95.4% of trials were good in the HNA group and 95.2% in the LNA group (good trials rates ranged from 90% to 100%). Bad channels were replaced by the interpolation of good channels in proximity to the bad ones according to the spherical spline algorithm (Srinivasan▇▇▇▇▇▇▇▇▇▇, ▇▇▇▇▇, ▇▇▇▇▇▇, ▇▇▇▇▇▇▇▇▇▇▇, & ▇▇▇▇▇▇▇, ▇▇▇▇1996). After bad channel replacement, all segments for each condition of each participant were averaged individually. Finally, ERPs were baseline corrected using a 200 ms pre-stimulus interval. The epoch length was 1200ms. The purpose of the post test, which took place immediately after ERP recording, was to classify images according to the source of name disagreement. Accordingly, only low name agreement pictures were used. Participants were instructed to press a key to indicate the relationship between a picture and a written name. In each trial, a low name agreement picture was presented with its name (according to ▇▇▇▇▇▇▇▇ et al., 1997), and two choices were given: (1) ―This is ONE OF the names of the object‖ (i.e., the disagreement source is alternative names for the object depicted (AN subgroup); or