Cognitive Neurology/Neuropsychology ©2010 – 2023




Our research has as its overarching goal the improvement of human cognitive function. This particular phase of our research program began in 1996, with an emphasis on understanding and treating the speech and language difficulties of nonverbal, low-functioning individuals with autism, and this remains a central focus. From the outset, however, we believed that insights and approaches gained from the study of speech and language could be applied to other cognitive processes. We saw that methods used for low-functioning individuals with autism could likely benefit persons with other conditions, both developmental and acquired, and even normally developing people. Further, we thought that studies of individuals with different conditions might synergize with each other. All this has proven to be true. So the Research Program now encompasses both developmental and acquired disorders, as well as the range of normal human variation.

We plan the research program in 5- and 10-year intervals, asking: What investigative, diagnostic, and treatment methods might be developed or prove effective over these spans of time? Therefore, while many of the issues that we study can be considered very basic ones, our efforts have necessarily been aimed at more translational and applied targets than would ordinarily be the case in a purely “basic science” approach.

The research program has also been shaped by the joint issues of safety and efficacy: Whatever is being considered, and whatever is being developed, must entail little or no risk to the participant. Whatever interventions are being considered must be of proven or potential benefit. Our focus on safety and efficacy is essential to justify trials on such participants as children and young adults. This has narrowed the types of potential interventions that we have elected to develop and test.

Speech, language, and their supporting cognitive abilities (such as working memory) are key components of human cognition. They are among the most needed abilities, and they are among the most affected abilities in nonverbal, low-functioning individuals with autism. Thus, for both practical and theoretical reasons, these cognitive processes have been central targets of our research. On a practical level, speech and language are vital for giving instruction and for understanding an individual’s needs. This is true whether the individual is nonverbal because of autism, or suffers from aphasia caused by a stroke or Alzheimer disease. On a theoretical level, there are strong reasons to believe that the cognitive abilities that are the foundations of speech and language are also foundations of human thought in general. Moreover, speech and language may themselves amplify other cognitive capabilities. For example, the very use of speech and language may help the mind “chunk” information and increase its capacity for conscious control.

Accordingly, we have sought to apply existing methods and develop novel approaches to assess and improve speech, language, and their related cognitive capabilities.

We must emphasize that all scientific efforts to study and treat individuals who have little or no language, whether due to autism or an acquired aphasia, are still in the relatively early stages compared to those for most other conditions. Attempts to understand the mental world of nonverbal individuals have been fraught with controversy, as have efforts to teach them speech and language skills. Moreover, it is extremely difficult to try to study and treat persons who do not understand ordinary means of communication, and whose overall behavior may be extremely challenging.

Despite these obstacles, our research can boast of two particularly noteworthy accomplishments over the past 5 years. First, through behavioral methods alone, we enabled a previously nonverbal adolescent with autism not only to develop increasing spontaneous speech, but apparently to use elements of syntax that he was never explicitly taught (Loughlin et al, 2011). The fact that this young man achieved oral speech at all is extraordinarily rare, if not unique (Pickett et al, 2009). If his speech truly uses combinatorial and inflectional syntax, as we believe it does, he is beginning to meet criteria for acquiring true language—as far as we know, an unprecedented achievement.

Our other notable recent accomplishment has been to study aggressively the use of transcranial direct current stimulation (tDCS) of the brain as a safe and potentially effective means of improving some cerebral functions. We have preliminary evidence supporting the hypothesis that tDCS can be effective in altering automatic lexical semantic processes and related aspects of cognitive control in normal individuals (Vannorsdall et al, 2012) and in individuals with schizophrenia and their first-degree relatives (van Steenburgh et al, 2012). Further, we are developing advanced simulation methods that will let us more precisely target the electrical effects of tDCS to specific neural structures (Sadleir et al, 2010; Sadleir et al, submitted). Once we can couple behavioral training methods with focused application of tDCS, we hope to be better able to improve cerebral functions across a wider range of conditions, as well as in normally developing individuals.

Scientific Background

Cognitive science and cognitive neuroscience have shown that overt human abilities are typically the product of multiple subsystems, some acting sequentially, some in parallel. These subsystems—and the neural processes that underlie them—can be viewed from a nearly overwhelming number of levels in space and time. But our goals allow a more manageable perspective. The processes most relevant to the abilities that we are targeting operate over time scales ranging from tens of milliseconds to seconds (except in the case of permanent learning), and over spatial scales in the nervous system that are on the order of millimeters, to at most centimeters.

While the dynamics of these systems are highly complex, we can characterize the peak ability of a given system at a given time as being limited by a particular subsystem (analogous to the rate-limiting step in a chemical reaction). It seems to be almost a truism, and it is certainly our assumption, that enhancing the performance of a function requires improving the performance of its rate-limiting subsystems. Even for the same overt function, the rate-limiting step is likely to be different for different conditions, and probably to some extent different even for different individuals. For example, for normally developing children who are learning to read, the usual rate-limiting steps are learning the visual forms of letters and words, and their relations to the sounds of the language. This is because most normally developing children already have excellent aural and oral vocabularies by the time they start learning to read. In contrast, minimally verbal individuals with autism have a limited aural/oral vocabulary. For these individuals, we still must determine what are the rate-limiting steps at any given moment, and how best to intervene.

To do so behaviorally, we have relied upon decades of research identifying the factors that have been proven to strengthen cognitive abilities: the person’s pre-existing level of ability, the degree to which a critical function can be engaged by the tasks used to teach it, the level of arousal and attentiveness to the process, the training intervals (e.g., massed versus spaced practice), and the total training time. These factors seem to interact multiplicatively, in almost the literal sense.

These general factors have been known empirically for a long time. Only in the past several decades, however, has science begun to understand the complex network of subfunctions that underlie overt behavior. Only recently have individual domains been better investigated, and the special requirements of some subsystems become evident, such as those involved in learning to speak versus those involved in learning to read printed words. Only recently have improved techniques been developed for identifying these subprocesses. This has been true at the functional (cognitive) level, through techniques developed in experimental psychology and cognitive science. It has also been true at the neurobiological level, through functional imaging of metabolic, structural, and electrical processes. Only very recently have techniques been developed that have the potential to manipulate cognitive processes on the scales of time and space commensurate with our target processes: techniques such as transcranial Direct Cortical Stimulation (tDCS), direct cortical electrical stimulation, and transcranial magnetic stimulation (TMS).


Domains of research

Our most active and best-developed investigations involve speech articulation (in people with developmental apraxia of speech), lexical semantics, cognitive control of lexical-semantic selection, and reading aloud and for meaning. Other areas of study include the neural processing deficits that might contribute to the intellectual impairment in many persons with autism; developmental deficits in speech perception, visual perception (early visual processing and the interactions that give rise to visual illusions), concept formation, and organization; network modeling of the semantic system; semantic generalization; elements of syntax; initiation of spontaneous behaviors; motivation and enjoyment in learning; and verbal and visual creativity.


We have studied normally developing adults, normally developing children, low- and high-functioning individuals with autism, individuals with aphasia caused by focal lesions or degenerative conditions, adults with amnesia, and normally aging elderly adults.

Assessment methods

To establish baseline levels of cognitive function and determine whether interventions enhance function, we have used standard behavioral response methods. We have also developed novel approaches to assessment. One important augmentation of standard methods has been extensive audio and video recording of the vocalizations and possible communicative efforts of prelinguistic children. All the behavioral measures have been interpreted with the logic of cognitive science and cognitive neuroscience.

We are also pursuing methods that do not require conscious behavioral responses: pupillometry, eye movement tracking, and EEG recordings of early- and late-evoked potential measures, as well as tDCS to induce temporary, reversible effects on cerebral functioning. (While tDCS is cited here as an assessment tool, it also shows promise in enhancing learning, and is discussed in more detail in “Intervention methods” below.) To our knowledge, we have the largest group of low-functioning individuals with autism ever studied with pupillometry, eye tracking, or EEG.

We have used assessment methods to study these cognitive functions (the assessment methods used for each function are shown in brackets [ ]):

  • Reading aloud [behavioral] (Ledoux & Gordon, 2011)
  • Vocabulary knowledge [pupillometry, eye tracking, event-related potentials] (Ledoux et al, 2009; Van Droof et al, 2010; Gangopadhyay et al, 2011)
  • Lexical ambiguity resolution [behavioral, event-related potentials]
  • Syntactic priming [behavioral, event-related potentials] (Brothers et al, 2011; Ledoux et al, 2011)
  • Perceptual capability [behavioral, event-related potentials] (Sung et al, 2011)

Intervention methods

We have chosen to study intervention methods based on their safety and on their proven or possible efficacy. We have used a varied mix of behavioral techniques drawn from the literature: errorless learning (Terrace, 1963), inclusive plans for generalization (Carter & Hotchkis, 2002), varying types of feedback (Maas et al, 2008; Hula et al, 2008), effects of over-learning (Rohrer & Taylor, 2005), and the method of least prompting (Horner & Keilitz, 1975). We have taken a particular interest in studying the comparative benefits of veridical repetition versus repetition with variation, in learning in patients with amnesia (Stark et al, 2005, 2006, 2007).

A major development in the past several years has been our addition of tDCS to behavioral training paradigms. Preliminary data show that tDCS is an effective tool for relatively focal stimulation of the cerebral cortex. We are investigating whether using tDCS in conjunction with focused behavioral training has synergistic effects. This combination is especially promising because tDCS has proven to be safe and unobtrusive, in sharp contrast to other putative methods of improving cerebral function. We have assessed the patterns of intracerebral current distribution that tDCS produces, to develop more targeted delivery of current to designated brain regions (Sadleir et al, 2010; Sadleir et al, submitted). We have also shown that tDCS can selectively improve (or impair) word retrieval in normal adults (Vannorsdall et al, 2012). Studies are underway of the effects of tDCS on word retrieval and other related functions in individuals with autism, stroke, and focal degenerative brain disease. To date (1 March 2013), we have studied a total of 196 participants with tDCS (174 normal adults, 11 with schizophrenia or 1st degrees relatives of individuals with schizophrenia, 3 with aphasia, and 8 high-functioning individuals with autism).

We have applied intervention methods in particular to these cognitive functions (the methods used for each are shown in brackets [ ]):

  • Speech articulation and spontaneity [behavioral]
  • Reading aloud for accuracy and meaning [behavioral]
  • Speech and visual symbol comprehension [behavioral]
  • Lexical-semantic retrieval [behavioral + tDCS]
  • Cognitive control [behavioral + tDCS]
  • Executive function, working memory, and psychomotor speed [behavioral + tDCS]


The issues in trying to improve speech, language, and their cognitive underpinnings, both in nonverbal individuals with autism, and in others with related or informative conditions or abilities, require coordinated efforts and insights from many neurocognitive domains. The domains that we are already studying and intend to keep studying are:

  • Speech perception
  • Speech production
  • Visual perception of linguistic and other stimuli, including bottom-up and top-down processing
  • Reading for meaning
  • Reading aloud
  • Lexical semantics and semantic organization; concept formation
  • Amodal (prelinguistic) conceptual abilities
  • Syntax in expression and comprehension
  • Working memory for linguistic materials
  • Interactions between systems to enhance performance, eg, feedback between lexical knowledge and grammar
  • Cognitive control, both automatic and conscious
  • Intentionality and conscious decision making
  • “Creativity” in linguistic and visual perception and in linguistic production
  • The roles of declarative and procedural memory processes in supporting and enhancing performance in all of the above functions

Similarly, we expect to keep developing the assessment methods that we have been using in our study population:

  • Standard voluntary behavioral response methods, augmented by intensive audio and video recording and event logging, including event logging in naturalistic settings
  • Pupillometry and eye tracking to study nonconscious behavioral responses
  • Early- and late-component evoked potentials (including N400) to study nonconscious physiologic responses

We will keep focusing on the three safe, effective methods that have the potential to benefit everyone in our study population:

  • Increasing interest and emotional engagement to enhance endogenous learning systems
  • Intensive behavioral training, including techniques suggested by recent studies of, eg, learning while playing video games
  • tDCS

We have 4 major goals for the next 5 years (ranked in increasing order of difficulty):

  1. Increase the momentum of our current research efforts, particularly the studies of treatment of speech articulation, speech production, visual processing, and reading, and in the therapeutic (treatment) efforts involving joint behavioral methods and tDCS.
  2. Expand the range of functions studied (tasks used), and the range of subject populations.
  3. Possibly venture into ‘new’ domains related to our primary mission. These would be domains that offer potential high yields, both for scientific knowledge and for potential therapeutic impact, even though they are high risk. Possibilities include: motor “intelligence” and motor learning, and various aspects of ‘creativity’ (here defined as novel connections or combinations).
  4. Make a major effort at a computational synthesis across the various domains we are investigating and the various levels at which they might be studied. This computational synthesis will not occur in theoretical isolation. We want it to be linked to empirical data, either that already available, or that which is potentially feasible to collect (eg, direct cortical recordings). We expect this process will force us to critically re-examine our assumptions and our data (as well as those in the literature), and identify critical gaps in theoretical understanding and in existing data.


Current support for our research comes from Department of Defense Contract W81XWH-10-1-0404 (Gordon, PI); the Wockenfuss Endowment and Fund; the Benjamin and Adith Miller Family Endowment for Aging, Alzheimer’s, and Autism; the Therapeutic Cognitive Neuroscience Professorship; and the Therapeutic Cognitive Neuroscience Fund. Prior support has come from NIH Grant R21-MH068522 (Gordon, PI), competitive grants from the National Alliance for Autism Research and from Cure Autism Now, and gifts from the Appel Family Foundation, the family of Bernard Gordon, the Nancy Lurie Marks Family Foundation, and the Marino Family Foundation.