Thresholds of visibility for masked lexical, non-lexical, and non-linguistic items in aphasia

• ¹ University of Washington, Speech and Hearing Sciences, United States

Introduction Visual masking of primes is a method used to tap into automatic processing while reducing or eliminating conscious processing of the primes. Masking is achieved by presenting primes very rapidly and preceding and/or following them with additional visual stimuli that interfere with conscious processing of the primes (Greenwald, Klinger & Liu, 1989; Forster, Mohan & Hector, 2003). A recent series of studies using masked priming with aphasia (Silkes, Dierkes & Kendall, 2012; Silkes & Rogers, 2012; Silkes, 2014) have suggested this approach may be useful for understanding and treating automatic spreading activation in the language system in aphasia treatment. Automatic spreading activation is the mechanism by which linguistic networks are thought to operate, both within interactive activation (IA; Dell, 1986) and parallel distributed processing (PDP; Nadeau, 2001) models of language. In IA models, automatic spreading activation is the mechanism by which units of representation at one level of processing activate representations at other levels. Impairment of automatic spreading activation may lead to deficits in activation transmission, involving inadequate spreading activation between levels of linguistic representation, and deficits in representation integrity, involving an inability of the system to maintain activation of representations long enough for them to be selected from among competitors (Schwartz, Saffran, Bloch & Dell, 1994). In the context of PDP models, spreading activation is inherent in the simultaneous recruitment of an entire representational network when any part of the network is activated. If automatic spreading activation is impaired, it would interfere with the system’s ability to co-activate all elements of a network in a timely fashion, leading to impaired ability to retrieve information that is otherwise adequately encoded in the system. A number of methodological questions remain, however, before masked priming can be used to understand the status of automatic spreading activation in aphasia (Silkes & Rogers, 2010). One of these is the question of how short the prime exposure duration needs to be for a person to be unable to consciously process what they have seen. In one previous study that attempted to set individual visibility thresholds, differences were seen between the range of thresholds for typical adults (10-20 ms) and for people with aphasia (PWA; 10-100 ms) on a task that required no linguistic processing; while the masked items presented were all words, there was no language processing required for this presence/absence judgment task. Given the non-linguistic nature of this task, the difference between groups was surprising, and led to questions about the mechanism that might account for the difference. If aphasia is language-specific, then differences between groups may indicate that language processing was occurring even though it was not required. If aphasia is not language-specific, however, and includes impairments in broader cognitive domains such as attention (Murray, 2002) or automatic spreading activation mechanisms (Ferrill, Love, Walenski & Shapiro, 2012; Silkes & Rogers, 2012), it is possible that differences between groups were due to overall slower processing in the individuals with aphasia regardless of domain (i.e., linguistic vs. non-linguistic). Aims This project explores visibility thresholds for both adults with aphasia and typical adults using masked lexical, linguistic but not lexical, and non-linguistic stimuli. We asked whether there is a difference between PWA and typical adults in the exposure durations at which masked real words (RW), non-linguistic strings (NL), and non-words (NW) are detected. Method Participants The control group comprised 11 typical adults, 54-76 years old (mean = 63.55), with scores ≥50/60 on the Boston Naming Test (BNT) (Kaplan, Goodglass & Weintraub, 1983) and ≥23/36 on Raven’s Coloured Progressive Matrices (RCPM) (Raven, 1976). At the time of writing this proposal, the experimental group comprises 12 adults with mild to severe aphasia, 46-72 years old (mean = 62.73) (see Table 1 for details). Additional participants are expected in this group prior to presentation at this conference. Stimuli Stimuli in the masked position of each trial were strings of 5-8 characters presented in gray font in a white rectangle against a black screen (see Figure 1). In the RW condition, these stimuli were real words, in all upper-case letters. In the NW condition, stimuli were strings of English upper-case consonants. In the NL condition, stimuli were non-linguistic symbols (from Wing Dings font). The control condition had nothing in the white rectangle. All other stimuli were white font against a black screen. Experimental protocol Participants were seated in front of a computer, told that they would see things flashing on the screen, and asked to determine if the white rectangle in each sequence was empty or had something in it. Participants pressed a green computer key if they saw something in the target position, and a red key if they did not. Each trial began with a fixation mark presented for 1000 ms, followed by a blank black screen for 500 ms (see Figure 2). The masking sequence then began, with three forward masks that were consecutive strings of eight non-alphabetic characters (e.g., &, #, @, ?) presented for 50 ms each. The target screen then appeared for the designated exposure duration for that list, followed immediately by a backward mask of another string of non-alphabetic characters for 50 ms, and then a string of lower case letters for 1000 ms. Finally, the participant saw a question mark presented on the screen, which was the indicator to make a response. Each run comprised 10 trials: five control (blank) trials and five with stimuli in the masked position, in random order. First, practice runs were conducted with the masked primes clearly visible, so that participants understood what they were looking for. The experimental runs then began, starting with 100 ms and decreasing by 10 ms after each run in which the participant achieved at least 70% accuracy on the 10 trials in that run. If a participant scored ≤60% accuracy on a run, that same exposure duration was repeated. Once a score of ≤60% was achieved on four consecutive runs at a given exposure duration, that was determined to be that participant’s threshold for that condition. Each condition (RL, NW, NL) was tested as a block, and the sequence of presentation of conditions was varied by participant according to a Latin square design. Results There was a significant difference between the control and experimental groups for the RW condition (t = -2.53, p = 0.02), with a mean threshold of 29.09 ms (sd = 15.14) for the control group and a mean threshold of 46.36 ms (sd = 16.9) for the experimental group. There was no significant difference between the groups for the NW and NL conditions (t = -0.88, p = .39, and t = -1.23, p = .23, respectively), although both conditions showed a trend toward the control group having lower thresholds than the experimental group (NW: 35.45, sd = 15.73 vs. 43.18, sd = 24.52; NL: 33.64, sd = 22.81 vs. 46.36, sd = 25.8) (see Table 2). Discussion Data collected to date demonstrate a clear difference between individuals with and without aphasia in their ability to perceive masked real words, but there appears to be no difference between groups for non-words and non-linguistic stimuli, although a trend is seen for these groups. Given the high variability for the NW and NL conditions, these analyses may be underpowered; therefore, data collection is ongoing and a clearer picture should be available by the time of presentation. Regardless of the eventual outcome, this poster will discuss the theoretical motivation for the study, and will discuss the possible implications for understanding the nature of underlying deficits in aphasia.