Knowledge and Learning of Verb Biases in Amnesia

Rachel Ryskin; Zhenghan Qi; Natalie Covington; Melissa Duff; Sarah Brown-Schmidt

doi:10.1016/j.bandl.2018.04.003

. Author manuscript; available in PMC: 2019 May 22.

Published in final edited form as: Brain Lang. 2018 May 22;180-182:62–83. doi: 10.1016/j.bandl.2018.04.003

Knowledge and Learning of Verb Biases in Amnesia

Rachel Ryskin ^1,², Zhenghan Qi ³, Natalie Covington ⁴, Melissa Duff ⁴, Sarah Brown-Schmidt ⁵

PMCID: PMC6048964 NIHMSID: NIHMS973366 PMID: 29775775

Abstract

Verb bias—the co-occurrence frequencies between a verb and the syntactic structures it may appear with—is a critical and reliable linguistic cue for online sentence processing. In particular, listeners use this information to disambiguate sentences with multiple potential syntactic parses (e.g., Feel the frog with the feather.). Further, listeners dynamically update their representations of specific verbs in the face of new evidence about verb-structure co-occurrence. Yet, little is known about the biological memory systems that support the use and dynamic updating of verb bias. We propose that hippocampal-dependent declarative (relational) memory represents a likely candidate system because it has been implicated in the flexible binding of relational co-occurrences and in statistical learning. We explore this question by testing patients with severe and selective deficits in declarative memory (anterograde amnesia), and demographically matched healthy participants, in their on-line interpretation of ambiguous sentences and the ability to update their verb bias with experience. We find that (1) patients and their healthy counterparts use existing verb bias to successfully interpret on-line ambiguity, however (2) unlike healthy young adults, neither group updated these biases in response to recent exposure. These findings demonstrate that using existing representations of verb bias does not necessitate involvement of the declarative memory system, but leave open the question of whether the ability to update representations of verb-specific biases requires hippocampal engagement.

Keywords: Hippocampal Amnesia, Language Processing, Eye-tracking, Syntactic Ambiguity

Listeners represent the co-occurrence frequencies between verbs and the syntactic structures they appear with—verb biases—and deploy this knowledge during online sentence processing. Words within sentences combine in different ways such that the prepositional phrase “with the feather” in a sentence like “Feel the frog with the feather” can attach to the noun, indicating which frog, or it can attach to the verb, indicating how to feel it. Knowledge about the statistical regularities with which particular verbs are used in one syntactic construction or another guides interpretation of these ambiguities (Snedeker & Trueswell, 2004; Garnsey, Pearlmutter, Meyers, & Lotocky, 1997). Moreover, this lexically-specific knowledge is updated based on recent linguistic experience (Coyle & Kaschak, 2008; Ryskin, Qi, Duff, & Brown-Schmidt, 2017a; 2017b; 2018).

In the present research, we probe the neural mechanisms that support the use of these statistical regularities in online syntactic processing. Our approach combines the study of neuropsychological patients with severe relational memory impairment, with psycholinguistic techniques to probe the online processing of sentences in rich contexts. In particular, we investigate a role for the hippocampal relational memory system, in light of its processing capabilities, in the online processing and updating of verb bias information. While previous work has demonstrated that offline identification of syntactic ambiguity is intact in patients with medial temporal lobe damage (e.g. Schmolck, Stefanacci, & Squire, 2000), emerging research examining language in rich contexts and with online measures points to a critical role for hippocampal relational memory in language processing (Duff & Brown-Schmidt, 2012; 2017).

Verb Bias Use in Language Processing

Language-wide distributional characteristics of specific verbs guide the online and offline resolution of prepositional phrase (PP) attachment ambiguities in sentences with globally ambiguous syntactic structures such as Feel the frog with the feather (Snedeker & Trueswell, 2004). In a norming study, Snedeker and Trueswell (2004) characterized a set of verbs based on the relative degree to which a sentence containing a specific verb and a with prepositional phrase is completed by a Modifier noun (e.g. choose the dog with the pointy ears) or an Instrument (e.g. tickle the teddy bear with the feather)—a difference in syntactic structure that also changes the meaning of the sentence. Verbs were then categorized as Modifier-biased (i.e., PPs tended to attach to the head noun), Instrument-biased (i.e., PPs tended to attach to the verb), or Equi-biased (verbs that were in-between). In a subsequent study using the visual-world eye-tracking technique (Tanenhaus, et al., 1995) to examine online processing, other participants listened to spoken instructions in which the verbs appeared in globally ambiguous syntactic constructions while viewing a scene with toys that afforded both an Instrument and a Modifier based interpretation. Snedeker and Trueswell (2004) found that listeners resolved these global ambiguities by relying on information about the verb’s lexical bias—the likelihood of the co-occurrence of the verb and each syntactic alternative that can be associated with it (see also Boland, 1997; Boland, Tanenhaus, Garnsey, & Carlson, 1995; Garnsey et al., 1997; MacDonald, Pearlmutter, & Seidenberg, 1994; Spivey & Tanenhaus, 1998; Spivey-Knowlton & Sedivy, 1995; Spivey-Knowlton, Trueswell & Tanenhaus, 1993; Taraban & McClelland, 1988; Trueswell, 1996; Trueswell, Tanenhaus, & Kello, 1993; Wilson & Garnsey, 2009).

These findings point to a critical role for knowledge of the mapping between verbs and structures in on-line language processing. Here, we explore whether a brain structure better known for its role in memory – the hippocampus – may play a central role in online processing of verb bias information. The hippocampus has long been known to play an important role in relational memory. It has been implicated in encoding the enduring representations of the co-occurrences of people, places, and things as well as their spatial and temporal relations (Cohen & Eichenbaum, 1993; Ryan, Althoff, Whitlow, & Cohen, 2000; Eichenbaum & Cohen, 2001; Davachi, 2006). The hippocampus has also been tied to the fast, on-line binding of relational information between co-occurring stimuli (Hannula, Tranel & Cohen, 2006; Barense, Gaffan, & Graham, 2007; Ranganath and D’Esposito, 2001), and to memory-perception comparisons (e.g., Duncan, Curtis, & Davachi, 2009; Harrison, Duggins, & Friston, 2006; Kumaran & Maguire, 2007; Kim, Lewis-Peacock, Norman, Turk-Browne, 2014; Long, Lee, & Kuhl, 2016). Further, recent work has suggested an important role for the hippocampus in aspects of language processing with similar processing demands to online syntactic disambiguation (Kurczek, Brown-Schmidt, & Duff, 2013; Rubin, Brown-Schmidt, Duff, Tranel, & Cohen, 2011; Piai et al., 2016; Blank, et al., under review). For instance, patients with hippocampal amnesia struggle to resolve ambiguity during online sentence comprehension when interpretation hinges on binding an ambiguous pronoun to the appropriate antecedent (e.g., Melissa is playing violin for Debbie as the sun is shining overhead. She is wearing a blue dress). The online processing of verb bias information may similarly engage hippocampal relational binding mechanisms as it requires in-the-moment retrieval of verb-specific information and binding of the prepositional phrase to what it modifies.

Alternatively, other evidence suggests that many aspects of syntactic processing do not rely on the hippocampal declarative memory system. For example, artificial grammar learning abilities and susceptibility to syntactic priming remain intact in patients with amnesia (Ferreira, Bock, Wilson, & Cohen, 2008; Schmolck et al., 2000; Knowlton, Ramus, & Squire, 1992; cf. Chang, et al., 2012; MacKay, Stewart, & Burke, 1998). These findings are often taken as evidence that artificial grammar learning and syntactic priming tap procedural memory mechanisms. Thus, whether processing of verb bias takes place within, or in concert with, the hippocampal declarative memory system is an open question.

Further, verb bias information is not static, but rather is malleable based upon experience. Previous research has shown that experience with a particular syntactic structure facilitates processing of and even expectation for that structure in the future (Fine & Jaeger, 2013; Fine, Jaeger, Farmer, & Qian, 2013; Kaschak & Glenberg, 2004; Wells, et al., 2009; Luka & Barsalou, 2005; Tooley, Traxler, & Swaab, 2009; Tooley et al., 2014). Moreover, listeners’ preferences for lexeme-specific syntactic attachment can be shaped by recent experience (Chang, Dell, & Bock, 2006; Chang, Janciauskas, & Fitz, 2012; Jaeger & Snider, 2013). Ryskin, et al., (2017a; 2017b; 2018) demonstrated that representations of the biases of specific verbs are malleable and can be updated through exposure to new structure-verb co-occurrence statistics. The use and updating of verb-structure relations based on recent linguistic experience requires the tracking and rapid updating of arbitrary co-occurrence information, thus we postulate that these mechanisms may place key demands on the hippocampal memory system.

A role for the hippocampus in tracking and updating verb bias information based on experience would be consistent with recent evidence of hippocampal involvement in statistical learning (see Schapiro, Turk-Browne, Botvinick & Norman, 2017). Neuroimaging data point to a role for hippocampus in the representation of statistical regularities in healthy adults (e.g. Turk-Browne, Scholl, Chun, & Johnson, 2008; Schapiro, Kustner, & Turk-Browne, 2012). Further, data from one patient with complete bilateral hippocampal loss indicate a failure to learn patterns of temporal co-occurrence in a variety of stimuli (Schapiro, Gregory, Landau, McCloskey, & Turk-Browne, 2014; cf. Covington, Brown-Schmidt, & Duff, in press). On the other hand, patients with hippocampal damage have been shown to demonstrate learning of patterns in linguistic material, such as artificial grammar learning (Knowlton et al., 1992), syntactic priming (Ferreira et al., 2008), and new dialectal variants tied to a particular speaker (Trude, Duff, & Brown-Schmidt, 2014). Thus, whether the hippocampus is critical in supporting the tracking of co-occurrences of verbs with syntactic structures and learning of new verb-biases is an open question.

In the present work, we take a hybrid neuropsychological – psycholinguistics approach to examining the biological memory systems that support the real-time processing of lexical biases in syntactic ambiguity resolution. In Experiments 1 and 2, we used paradigms modeled on Snedeker & Trueswell (2004), with real-world objects and a computerized paradigm respectively, to examine the use of verb bias in patients with hippocampal lesions as well as healthy, demographically-matched comparison participants. To preview, patients with amnesia, as well as older adult comparisons made use of verb bias information in order to resolve global syntactic ambiguity in the moment. In Experiment 3, we aimed to test whether patients with amnesia can not only use verb bias information online but also update those biases based on recent exposure. We employed a paradigm we have successfully used in the past to demonstrate dynamic updating and use of verb biases in healthy young adults (Ryskin, et al., 2017a; 2017b; 2018). Amnesic and comparison participants were exposed to sets of—initially equi-biased—verbs that were exclusively tied to one of two possible syntactic structures (either instrument or modifier-interpreted with phrases) by a disambiguating context. We then measured offline behavioral responses and eye-fixations on globally ambiguous test trials to evaluate whether the listeners’ interpretation of a specific verb-argument structure was influenced by the syntactic structures that the verb had been paired with previously.

General Methods

Across the three experiments, participants included 5 individuals with bilateral hippocampal damage (one female) and severe declarative memory impairment and 8 healthy comparison participants. Etiologies of the patients with hippocampal amnesia included anoxia/hypoxia (1846, 2363, 2563) resulting in bilateral hippocampal damage, and herpes simplex encephalitis (HSE) (1951, 2308), resulting in more extensive bilateral medial temporal lobe damage affecting the hippocampus, amygdala, and surrounding cortices. Structural MRI data were available for 4 of the 5 patients (excluding 2563) (Figure 1) and volumetric analyses performed on 3 of the 5 patients revealed significant reduction to the hippocampus bilaterally with volumes reduced by at least 2.64 studentized residuals compared to age matched healthy comparison participants (Allen et al., 2006). Visual inspection of a CT scan from patient 2563, who wears a pacemaker, confirmed damage limited to hippocampus.

Magnetic resonance scans of hippocampal patients. Images are coronal slices through four points along the hippocampus from T1-weighted scans. R = right; L = left; A = anterior; P = posterior; NC = a healthy comparison brain.

Performance on tests of neuropsychological functioning revealed a severe and selective impairment in episodic/declarative memory functioning while performance across other cognitive domains was within normal limits (Table 1). The Wechsler Memory Scale–III General Memory Index scores for each participant were at least 25 points lower than their scores on the Wechsler Adult Intelligence Scale–III, and the mean difference between Full Scale IQ and General Memory Index was 38.5 points. The average Delayed Memory Index was 61.4, which is 3 standard deviations below population means. This deficit in episodic/declarative memory was observed in the context of otherwise intact cognitive abilities. Participants performed within normal limits on standardized neuropsychological tests of intelligence, language, and executive function.

Table 1.

Demographic and neuropsychological characteristics of patients with amnesia

Subject	Expt.	Sex	Birth Year	Hand.	Ed.	Etiology	HC Damage	HC Volume	WAIS III FSIQ	WMS III GMI	BNT	TT	WCS (Cat)
1846	1, 2, 3	F	1963	R	14	Anoxia	Bilateral HC	−4.23	84	57	43	41	6
2363	1,2	M	1956	R	18	Anoxia	Bilateral HC	−2.64	98	73	58	44	6
2563	1,3	M	1955	L	16	Anoxia	Bilateral HC	N/A	102	75	52	44	6
1951	2,3	M	1952	R	16	HSE	Bilateral HC +MTL	−8.10	106	57	49	44	6
2308	1	M	1956	L	16	HSE	Bilateral HC +MTL	N/A	87	45	52	44	N/A
Group Summary		1F 4M		3R 2L	16 (±1.4)			−5.0 (± 2.8)	95.4 (±9.5)	59.0 (±12.2)	51.8 (±5.5)	43.4 (±1.3)	6

Open in a new tab

Expt. = which of the three Experiments this person participated in. HC = hippocampus (participants with bilateral hippocampal damage). Hand. = Handedness. Ed. = years of completed education. HSE = Herpes Simplex Encephalitis. + MTL = damage extending into the greater medial temporal lobes. N/A = no available data. Volumetric data are z-scores as measured through high resolution volumetric MRI and compared to a matched healthy comparison group (see Allen et al., 2006, Buchanan et al., 2005 for additional details). WAIS-III FSIQ = Wechsler Adult Intelligence Scale–III Full Scale Intelligence Quotient. WMS-III GMI = Wechsler Memory Scale–III General Memory Index. BNT = Boston Naming Test. TT = Token Test. WCS = Wisconsin Card Sorting Task; Cat = Number of categories achieved out of six.

Bolded scores are impaired as defined as 2 or more standard deviations below normative data.

Comparison participants were eight healthy individuals matched pairwise on age, handedness, sex, and education. All participants were native, monolingual speakers of North American English with normal or corrected-to-normal hearing and vision. Due to scheduling conflicts, not all patients with hippocampal amnesia participated in all studies. One healthy comparison participant completed all three of the experiments, three healthy comparisons participated in two of the three experiments, and one comparison only participated in the first experiment.

Experiment 1

The goal of Experiment 1 is to test the use of verb bias in patients with hippocampal amnesia using an experimental paradigm modeled on Snedeker and Trueswell (2004).

Participants

Patients 1846, 2308, 2363, and 2563 participated in Experiment 1 and were 55.5 years old and had 16.0 years of education, on average, at the time of data collection. The four demographically matched healthy comparison participants were 55.6 years old and had 17.3 years of education, on average.

Design & Procedure

Participants sat in front of an inclined podium with a shelf in each quadrant on which toys could be placed, and a central opening for a camera that recorded the participants’ eye-movements at 30 Hz. Another camera positioned behind the participant recorded their actions. The recorded instructions were played from a laptop computer through external speakers positioned at the top of the podium display. Participants were told that they would play a game involving following instructions. At the start of each trial, the experimenter placed the toys in each quadrant of the podium, introducing each by name as he did so. The mini-instruments attached to each animal were introduced as separate objects (e.g., “Here’s a crayon, a pig, a flower, another flower, an elephant, and another crayon”). After naming the objects in this manner a second time, the experimenter played the pre-recorded instructions for the trial. Each trial began with an instruction to "look at the camera", followed by two other instructions. In critical trials the first instruction was the critical sentence and the second was a filler sentence. This procedure is modeled closely on the one described by Snedeker and Trueswell (2004). Each participant heard all 24 critical sentences ordered pseudo-randomly so that no consecutive critical trials contained verbs with the same bias condition. Filler trials were randomly interspersed with the critical trials to mask the purpose of the experiment.

Each participant completed the same list of trials three times with different orders. As a result, each participant completed 72 critical trials. Testing was conducted over a series of 3 sessions, with all sessions for a given participant conducted within the span of a single week.

Materials

Both critical and filler trials included two instructions. In the critical trial, the first command is the critical sentence. 24 critical sentences and 72 filler sentences, all from the materials of Snedeker and Trueswell (2004), were recorded by a native English female speaker. Each critical sentence contained a target verb with either Instrument Bias, Equi Bias or Modifier Bias, followed by an animal name in direct object position, and an ambiguous with phrase, as in examples (1–3). The average critical sentence duration was 2.15s and there was no significant difference among three bias conditions in duration (F(2,21)=2.56, p = 0.101). The filler sentences are unambiguous instructions using different verbs.

Tickle the frog with the feather. (Instrument Bias)
Point at the pig with the flower. (Equi Bias)
Choose the cow with the stick. (Modifier Bias)

As shown in Figure 2, each critical sentence was accompanied by a set of toys: a Target Animal with a small replica of the target instrument (e.g., a pig holding a flower), a Distracter Animal with a small replica of the distracter instrument (e.g., an elephant holding a crayon), a Target Instrument (e.g., a large flower) and a Distracter Instrument (e.g., a large crayon). These toys made available both modifier and instrument interpretations of the with phrase. That is, the display contained a large flower with which to point at the pig, and a pig that had a flower. Due to our primary interest in the verb-bias manipulation, all critical sentences were presented in contexts comparable to the one-referent condition of Snedeker and Trueswell’s (2004) experiment (we did not test the two-referent condition).

Experiment 1 toy layout for the sentence *Point at the pig with the flower*. The target animal is the pig, which is holding a flower. The target instrument is the big flower.

Coding

Trained coders, blind to the hypotheses of the study, categorized participants’ actions as Instrument actions (the participant performed the action on the target animal using the target instrument), Modifier actions (the participant performed the action on the target animal using her hand), or Mini-instrument actions (the participant used the mini-instrument attached to the target animal to carry out the action). Because the mini-instrument actions are potentially consistent with both an NP-attachment and a VP-attachment interpretation, six trials (5 in the patient group and 1 in the control group) coded as Mini-instrument actions were removed from the eye-movement analysis as in Snedeker & Trueswell (2004). Four additional trials in the patient group were excluded from analysis because the participant did not act (2) or because the participant asked where the target animal was (1) or because the participant was out of the camera view (1).

Coders first marked the onset of each critical sentence and then participants’ eye movements were coded frame by frame from video, played back without sound. Coders recorded where the participant was looking from the onset of the critical sentence until 3.2 seconds later. We coded looks to the four quadrants of the podium, away, and to the central camera. Frames were coded as missing if the participants’ eyes were hidden. Four trials were removed from analysis, because participants made fixations to a quadrant of the podium in fewer than 1/3 of frames between the onset of the sentence and the average onset of the action (70 frames). As a result, of 576 possible trials, we eliminated a total of 39 trials (6.8%) from the eye-movement analyses.

Results

Actions

The participants’ responses to the instructions were coded as Instrument or Modifier actions as described above. Participants in both groups were more likely to perform an Instrument action in the instrument-biased condition than the equi-biased or modifier-biased conditions (Figure 3).

All analyses were performed using the lme4 package in R (Bates, Maechler, Bolker, & Walker, 2015). The action data were analyzed using mixed-effects logistic regression. The dependent measure was whether the participant made an Instrument response or a Modifier response. The model included random intercepts for participants and items, as well as random by-participant slopes for verb bias condition and random by-item slopes for participant group. Verb bias condition was entered as a pair of orthogonal contrast codes. The first contrast compared the instrument-biased condition (−0.66) to the average of the equi-biased (0.33) and modifier-biased (0.33) conditions. The second contrast compared the equi-biased (−0.50) condition to the modifier-biased (0.50) condition. Participant group was deviation coded (Comparisons: −0.5, Amnesics: 0.5). The full set of parameter estimates is reported in Appendix Table A1. Instrument responses were significantly less likely (b=−3.34, SE=0.76, z=−4.42, p<0.001) in the equi-biased and modifier-biased conditions compared to the instrument-biased condition. There was no significant difference in participant actions between the modifier-biased condition and the equi-biased condition (b=0.74, SE=0.97, z=0.77, p=0.44). There was no main effect of participant group (b=−0.63, SE=1.34, z=−0.47, p=0.64) and verb bias condition did not interact with participant group (Contrast 1: b=1.15, SE=1.05, z=1.10, p=0.27; Contrast 2: b=0.79, SE=1.23, z=0.64, p=0.52).

Eye fixations

The plot of fixations over time (Figure 4) suggests a similar pattern of results to Snedeker and Trueswell (2004) in the amnesic group. The amnesic participants seem to make more fixations to the Target Animal when the verb is modifier-biased than when it is equi-biased, and the instrument-biased verbs elicit the fewest fixations to the Target Animal. The reverse pattern is observed for fixations to the Target Instrument. This pattern is numerically weaker in the comparison participants.

Experiment 1: Proportion of gaze fixations in three conditions (Modifier-biased, Equi-biased, and Instrument-biased) over time for persons with amnesia and demographically-matched comparison participants (graphs for each individual available in Supplemental Material S2). (Dashed lines indicate average beginnings of the three time windows.)

We analyzed the data in three time windows in order to capture eye-movements elicited during both early and later portions of the sentence. All time windows were offset by 200 milliseconds to account for the time needed to program and launch an eye movement (Hallett, 1986). The first time window started at the onset of the verb (e.g., Point at the pig with the flower) and ended at the onset of the first noun (e.g., pig). The average duration of the first time window was 659 milliseconds. The second time window began at the onset of the first noun and ended at the onset of the second noun (e.g., flower). The average duration of the second time window was 808 milliseconds. The third time window started at the onset of the second noun and ended 1500 milliseconds later (the maximum duration of the second noun was 852 milliseconds).

We fit mixed-effects linear regression models predicting eye fixations from participant group and verb bias condition, across the three time windows of interest. The dependent variable for these analyses was a measure of Target Animal Advantage: proportion of fixations to the target animal minus the proportion of fixations to the target instrument, with an empirical logit transformation applied to the proportions because the data were not normally distributed (see Barr, 2008 for a related approach). The model included random intercepts for participants and items, as well as random by-participant slopes for verb bias condition, time window, and their interaction and random by-item slopes for participant group, time window, and their interaction (this was the maximal random effect structure justified by the design). Verb bias condition was entered as a pair of weighted¹ orthogonal contrast codes. The first contrast compared the instrument-biased condition (−0.66) to the average of the equi-biased (0.34) and modifier-biased (0.34) conditions. The second contrast compared the equi-biased (−0.50) condition to the modifier-biased (0.49) condition. Participant group was deviation coded (Comparisons: −0.5, Amnesics: 0.5). Time window was Helmert coded (Contrast 1: Time Window 1= −0.66; Time Window 2 = 0.33, Time Window 3 = 0.33; Contrast 2: Time Window 1= 0; Time Window 2 = −0.5, Time Window 3 = 0.5). All analyses were performed using the lme4 package in R (Bates, Maechler, Bolker, & Walker, 2015) with p-values estimated using Satterthwaite’s approximation performed by the lmerTest package (Kuznetsova, Brockhoff, & Christensen, 2015). The full set of estimated parameters from this model is summarized in Appendix Table A2. There was a significant effect of Time Window, such that Target animal advantage was larger in the second and third time windows relative to the first (b=2.00, SE=0.58, t=3.44, p=0.003). There were no main effects of participant group (b=0.79, SE=0.62, t=1.28, p=0.25) or verb bias condition (Contrast 1: b=0.53, SE=0.78, t=0.67, p=0.51; Contrast 2: b=−0.03, SE=0.91, t=−0.03, p=0.97), nor did these significantly interact with time window.

Discussion

The data from this experiment provide some evidence that patients with amnesia use verb bias information while processing global syntactic ambiguities as measured by their offline, behavioral responses. The statistical analysis of the eye-movements does not reveal significant differences across conditions. Yet, the apparent lack of such an effect in the comparison group alongside a trend towards a verb bias effect for the patients, points to potentially noisy data and a low-power design. Similarly, the action data do reveal a significant difference between the Instrument-biased condition and the Equi-biased and Modifier-biased conditions, consistent with Snedeker and Trueswell (2004), but no difference between the Equi-biased and Modifier-biased conditions. Thus, these data suggest that the current design may not have had enough power to robustly detect the effects of verb bias. These observations suggest that a replication with a larger number of trials is necessary to draw strong conclusions about the role of verb bias in language processing for patients with hippocampal damage and impairments in declarative (relational) memory.

Experiment 2

The goal of Experiment 2 was to test the use of verb bias information in patients with amnesia and healthy comparison participants using a more highly powered design. While the use of real objects and a physical display in Experiment 1 allowed for a close replication of Snedeker and Trueswell (2004) in this population, in Experiment 2 we leveraged the efficiency of a computer-based display to substantially increase the number of trials.