Workshops

Morning Session

(9:30 am tp 12:30 pm)

Signal detection theory and consciousness research

By William P. Banks (WPB04747@pomona.edu)

I propose a two-part workshop on signal detection theory, three hours long, for the ASSC7 conference.

The first part will cover the basics of signal detection theory. I will present several types of threshold models and "correction for guessing," as well as their use in multinomial models. The receiver operating characteristic and its construction from confidence ratings will be presented, along with limitations on inferences that can be derived from it. The Gaussian class of model of detection theory will be presented, as well as techniques for deriving Gaussian parameters from data, with examples in Excel spreadsheets that can be used to analyze data. Given the amount of time available, this will not be a rigorous mathematical treatment but rather an intuitive approach, with the emphasis on understanding and using the concepts of detection theory.

The second part will show how these techniques can be applied to data sets in which questions about awareness are central. Implicit memory is one area in which detection measures can be applied, and some applications will be discussed. The issue of implicit memory as a component of normal memory, and of such unconscious components as "false fame," will be discussed. Multidimensional signal-detection models as appropriate to these cases and will be presented. Excel formulas and techniques will be demonstrated and made available to the attendees.

Time will be set aside during the second part for attendees to present their problems. I have been working for years on ways to analyze data with detection theory, and I will do my best to find the best application for your problem during the workshop, time permitting, or afterward. Attendees who have problems on which they would like advice are encouraged to send them to me before the workshop -- the more lead time the better -- and I will address at least some of them during the workshop, possibly using some as examples of applications.

 

Do Current Models of Color Perception Neglect Phenomenology?

By James A. Schirillo (schirija@wfu.edu) & Michael E. Rudd (mrudd@u.washington.edu)
Department of Psychology, Wake Forest University & Department of Psychology, University of Washington

Many consider visual awareness to be one of the best understood aspects of human consciousness, and color perception as one of the most thoroughly studied topics of visual awareness. There is also a longstanding interest in the pivotal role color perception plays within philosophy. The goal of this workshop is to both survey and critique the current state of research on color vision. We will argue that, while the field of color perception may be one of the most advanced in terms of both techniques and models, the field is still in its infancy in terms of its ability to characterize adequately the richness and multidimensional qualities of color in phenomenological awareness. Toward that end, we will discuss the following topics, which are all active areas of research within color science and philosophy.

1. Reflectance and illumination

A. The eye does not directly receive information about either the absolute intensity or spectral characteristics of the ambient illumination (e.g., sunlight versus fluorescent lighting). Nor does it receive direct information about the spectral reflectance properties of surfaces (e.g., their paint jobs or coloration). The light arriving at the retina confounds these two types of information. We will discuss how, and how well, the brain disentangles information about illumination and reflectance. Color theorists have constructed clever computational models to show how to achieve such disentanglement given reasonable assumptions, but these models are still far from perfect. They generally cannot account for many well-known color illusions and, moreover, the mistakes that the models make in judging object characteristics are not necessarily the same mistakes that people make.

People 'consciously perceive' 3-D 'empty' space as containing illumination (e.g., the room you are reading this abstract in is full of light), yet researchers in the field of visual perception never discuss this. Instead, they generally assume that humans are built to see objects, and consequently, that the visual system cares only about reflectance. They often view color itself as the subjective correlate of reflectance, and not of illumination, which we "discount." However, while people may be able to discard properties of the illuminant to deduce reflectance, the illuminant remains as shadows, etc. Moreover, the chromaticity and amount of light contained within a given volume of space is available to consciousness, which is what allows humans to decide if they should venture into any given cave.

B. In experiments performed with even moderately complex spatial patterns, such as Edwin Land's Mondrian stimuli, the qualia of "lightness" and "brightness," defined as the subjective magnitudes of "reflectance" and "physical intensity" respectively, behave differently. Yet in studies carried out with simple stimuli, "lightness" and "brightness" percepts are ambiguous. The field does a terrible job of distinguishing between lightness and brightness. We will discuss some of the problems involved in disambiguating these two perceptual dimensions.

2. Trichromacy

This issue divides the vision community into two camps: "color-channel theorists," whose approach is based on neural mechanisms, and "color-appearance theorists," whose approach is based primarily on phenomenology. Unfortunately, there is little cross-talk between these two communities. It is easy to show why these two camps exist, and we will discuss why these two approaches have failed to merge. We will contend that for the field to progress it is a paramount and necessary requirement to construct a bridge between these two approaches.

3. Edge integration

Many computational lightness models assume that reflectance is "reconstructed" by the brain through a process that involves first measuring the luminance ratios at all of the borders within a scene, then combining these ratios across space to obtain a scale of relative reflectances. This is, for example, the basic principle behind Land's well-known Retinex model of color vision. Two problems that arise in such models are: 1) correctly categorizing illumination and reflectance edges; and 2) anchoring the scale of relative reflectances computed by this procedure to absolute standards of lightness. We will review some of the current models that take this approach, all of which have limited applicability.

4. Lightness anchoring

A. Vision researchers have proposed several hypotheses to suggest how to anchor the reflectance scale to an absolute lightness scale. These include the "average luminance is gray" rule (otherwise known as the "gray world hypothesis") and the "highest luminance is white" rule. Many take the results of several recent studies as support for the latter idea, but the overall pattern of results obtained in anchoring studies is in fact complex. We will discuss a few studies that support the "gray world hypothesis" and some studies that have demonstrated effects on lightness by manipulating the luminances of dark regions within the image, which is inconsistent with the highest luminance rule. Furthermore, we will illustrate how the gray world hypothesis might do a better job of accounting for adaptation, something neural-channel theorists consider but phenomenologists often neglect.

B. Another difficulty for the theory of anchoring is the problem of "self-luminosity." Sometimes, but not always, the highest luminance, rather than appearing white (i.e., as a highly reflective surface), appears self-luminous (i.e., like a glowing object). We will consider arguments that place the "degrees of self-luminosity" on the same magnitude scale as lightness. However, this raises the conundrum that lightness and luminosity judgments correspond to fundamentally different qualia. Furthermore, we will elucidate the ramifications of not having a principled theory to explain the conditions under which we perceive self-luminosity.

5. Color transparency

The problem of color transparency is particularly vexing for color theory, and is the subject of much recent empirical research. For example, unlike illumination, filters both reflect light and undergo inner transmittance. Thus, when color filters are placed in front of a background field, the visual system must decide what properties of the local luminance (i.e., light hitting the eye) are due to the transmittance of the filter (i.e., both its opacity and spectral properties), which are due to the properties of the background, and which are due to the properties of the illumination. To decide which attributes of the retinal image belong to the filter, background, and illumination, requires that the brain construct a complex internal representation of the world, involving colored layers that appear at different depths with respect to the observer. We will discuss some recent theoretical approaches to this problem that look promising.

6. Lightness as a 1D color system

Experiments using achromatic stimuli (i.e., ones that exist on the gray scale, ranging from black to white) tend to produce data that is much more systematic than that produced by comparable chromatic experiments. But vision scientists often study lightness perception (i.e., achromatic color vision) under the assumption that it is simply a one dimensional system for developing models of color vision, more generally. The standard claim is that we want to build from the simple to the complex, and that understanding lightness/brightness phenomena will provide the fundamental laws that will ultimately generalize to trichromatic color phenomena. We will argue that the simple-complex metaphor of lightness and color is inaccurate, and that lightness perception and trichromatic color perception are more like apples and oranges. Thus, we will conclude with a discussion as to why those interested in color perception should reallocate their efforts to understand better the relationship between lightness and color vision, since they often fail to produce comparable results.

 

In search of NCC: Insights from single neuron recordings in the human brain

By Gabriel Krieman (kreiman@mit.edu)
Whiteman Fellow, Center for Computational and Biological Learning
Massachusetts Institute of Technology

A large fraction of the electrophysiological investigations about consciousness at the single neuron level comes from experiments performed in animal models, mostly in monkeys, cats and mice. It is possible to record the neuronal activity of individual neurons in the human brain in epileptic patients where electrodes are implanted in order to localize the seizure onset focus. Here we will discuss the methodology and compare the technique with other experimental tools used in the exploration of correlates of consciousness. We will describe some of the results that have been obtained by monitoring the neuronal activity in the human brain, focusing on the search for the neuronal correlates of consciousness in situations where the visual input is dissociated from perception. We will also give an overview of electrical stimulation in monkeys and humans. Finally, I will discuss the implications and possibilities for the scientific study of consciousness.

Overview:

PART 1: HISTORY AND METHODOLOGY OF SINGLE NEURON RECORDINGS IN HUMANS - THE MEDIAL TEMPORAL LOBE AND CONSCIOUSNESS.

1. Brief history of single neuron recordings (in humans and monkeys)
2. Difficulties and advantages of recording from humans
3. Comparison with other methodologies including fMRI, scalp EEG and others.
4. Anatomy of the monkey medial temporal lobe
5. The medial temporal lobe and consciousness
6. General characteristics of recordings in the human medial temporal lobe - Comparison with recordings in animal models including mice and monkeys.
7. Overview of human single neuron recordings related to language and memory:
    a. Single neuron activity and language
    b. Single neuron activity and memory
    c. Other single neuron work

PART 2: IN SEARCH OF NEURAL CORRELATES OF VISUAL CONSCIOSNESS

8. Visual responses in the human medial temporal lobe. Evidence for a sparce and explicit representation.
9. Visual imagery
10. Binocular rivalry and flash suppression

PART 3: Implications and possibilities for the scientific study of consciousness

11. Overview of electrical stimulation in monkeys and humans
12. Relevance, implications and possibilities for the scientific study of consciousness
13. Future directions

 

Complexity and consciousness

By Wei Ji Ma (ma@klab.caltech.edu)

Tononi and Edelman have suggested that a certain complexity measure applied to the brain can account for key properties of conscious experience [1]. Meanwhile, Koch and Laurent argue that a science of "brain complexity" should reflect structure, function, evolutionary history, and information coding schemes of nervous systems [2]. Can a single information-theoretical quantity accomplish this? The first obstacle is that there is no accepted universal definition of complexity, while many of the published measures are seriously flawed. The second issue is whether the neural correlates of any form of consciousness share properties which allow characterization by a global, quantitative marker like complexity. In this tutorial, we will first formulate principles for complexity and critically examine and categorize existing measures, such as predictive information in time series [3]. Next, we will ask ourselves what complexity means in biological systems. It has been hypothesized that complexity correlates with degree of adaptation [4] or with robustness [5], and is therefore selected for. However, can the concept of complexity cover anything beyond more traditional approaches? Finally, we will discuss whether our knowledge about neural mechanisms underlying perception, such as an elaborate hierarchical organization with recurrent interactions, or about the brain as a computational machine, justifies statements about the role of complexity in consciousness, and what this role might be: cause, consequence, byproduct, or equivalent?

References:

[1] G. Tononi and G.M. Edelman, Consciousness and Complexity, Science 282 (1998), 1846-1851
[2] C. Koch and G. Laurent, Complexity and the Nervous System, Science 284 (1999), 96-98
[3] W. Bialek, I. Nemenman, and N. Tishby, Predictability, Complexity, and Learning, Neural Computation 13 (2001), 2409-2463
[4] C. Adami, C. Ofria, T.C. Collier, Evolution of Biological Complexity, PNAS 97 (2000), 4463-4468
[5] M.E. Csete and J. Doyle, Reverse Engineering of Biological Complexity, Science 295 (2002), 1664-1669

 

Functional imaging in the conscious resting state, sleep, general anesthesia and coma

By Steven Laureys (steven.laureys@ulg.ac.be)
Department of Neurology and Cyclotron Research Center,
University of Liege, Liege, Belgium

Consciousness has two major components: arousal and awareness. We will review the latest functional imaging (PET, fMRI, and EEG) studies from our own and other laboratories on the conscious resting state, deep sleep (slow wave sleep), paradoxical (rapid eye movement) sleep, general anesthesia (i.e., pharmacological coma) and different states of (seemingly) altered consciousness encountered after severe brain pathology. Coma is defined as the absence of arousal and hence of awareness. The vegetative state is a unique condition wherein all aspects of awareness are abolished whereas arousal is preserved. Patients in a minimally conscious state may show limited emotional or motor behavior that is contingent upon the presence of eliciting stimuli (e.g., crying precipitated by particular family voices). The defining feature of the locked-in syndrome is the complete preservation of cognition in a completely paralyzed body (small blinks or vertical eye movements are the only way to communicate). Functional neuro-imaging permits to objectively measure how deviant from normal is the cerebral activity and its regional distribution, at rest and under various conditions of stimulation (e.g., auditory, noxious or emotional processing) in each of these different physiological, pharmacological and pathological states of altered consciousness.

 

Afternoon session

(2.00 pm to 5.00 pm)

Temporal and Spatial Analysis of Electroencephalographic Signals

By Walter Freeman (wfreeman@socrates.Berkeley.EDU) & Rodrigo Quiroga (rodri@vis.caltech.edu)

Consciousness for most of us is realized in a rapid sequence of thoughts, feelings, perceptions and mental images. Instead of seeking measures of the state or level of consciousness in subjects, as in evaluating surgical anesthesia or sleepiness, we propose to make images of the neural activity patterns in subjects who give verbal, instrumental or behavioral feedback on their mental states. Making such images is a challenging task that can be addressed with recent advances in EEG analysis. The objective of this tutorial is to describe these advanced methods of analysis of single and multi-channel electroencephalogram (EEG) recordings. These new algorithms can be applied in any clinical facility with standard equipment at no great expense. They have applications to the study of attention, intention, expectancy, sensory processing, formation of meaning, learning, habituation, sensitization, epilepsy, etc.

In the first part of the tutorial we will review conventional as well as advanced methods of analysis of EEGs in the temporal and frequency domains. One of the most distinctive features of the EEGs is the appearance of oscillations in different frequency bands, which reflect the synchronized activity of a large group of neurons. Brain oscillations have been correlated with different brain processes and its power is usually quantified by means of the Fourier Transform. The Fourier Transform is so far the most used tool for the analysis of EEGs, but it assumes stationarity of the signal and it does not give any time information. It is therefore not appropriate when frequency patterns change over time. For these cases, 'time-frequency' representations such as the one given by the Short Time Fourier Transform are more suitable. In particular, we will describe a new time-frequency decomposition, namely, the Wavelet Transform, and stress its advantages in the analysis of EEG data. Brain processes involving larger neuronal assemblies or interactions between distant sites are represented in correlations between EEG electrodes. In this respect, we will describe recently proposed measures of synchronization and compare them to conventional approaches.

In the second part we will review recent advances in EEG spatial pattern imaging, with emphasis on techniques for analysis of multi-channel scalp recordings from normal human volunteers. We will begin by describing the advantages of spatial analysis with 1-D arrays preparatory to 2-D recording. We will describe reformulation of the 1-D FFT for display in log-log coordinates. This display is useful to distinguish among various noise spectra and the "1/f" scaling that distinguishes EEGs from muscle potentials [electromyograms, EMGs]. We will show the application of the 1-D FFT to EEGs from curvilinear scalp electrode arrays to discuss temporal and spatial sampling, aliasing, the Nyquist frequencies, and the spectral distortions caused by the impedance barriers of the scalp and skull, and the sulci and gyri of cortex. We will demonstrate the subdivision of the temporal and spatial spectra
into the classical ranges by use of temporal and spatial band pass filters. We will discuss in detail a form of nonstationarity in brain dynamics, in which cortical states occur as brief stable EEG amplitude patterns, like frames in a movie film. Each window is bracketed by sudden changes in EEG phase patterns. We will introduce the Hilbert transform, as a complement to the Fourier transform, in order to get the high temporal resolution needed to document the phase jumps. We will conclude with discussion of the criteria for the temporal and spatial filtering that is necessary for effective use of the Hilbert transform.

 

Visibility, Brightness, Shape, and Their Relationship to the Neural Code

By Susana Martinez-Conde & Stephen L. Macknik (macknik@neuralcorrelate.com)
Institute of Ophthalmology, University College London

This tutorial will highlight studies of the underlying neural code of visibility and brightness perception, and its entangled relationship with, surprisingly, the fundamental neural basis of primary shape perception. These studies used a combination psychophysical, electrophysiological, optical imaging, and computational experimental methods.

The tutorial will describe the basic functional anatomy of the early visual system and address what we know about several fundamental questions concerning visibility: What is the physical substrate of the perception of visibility? What are the features of the world that are most visible and most readily sampled by the visual system? What are the neural correlates of these basic visible features? What parts of the brain must be stimulated with this activity for us to aware of the visible stimulus?

But seeing is more than perceiving that a stimulus is visible; it also requires that the stimulus's shape is discernible from the background and is recognizable. So we will also discuss fundamental questions of shape perception as well: Are the neurons of the early visual system truly edge detectors? If not, what features are most readily sampled by the early visual system?

The tutorial will feature demonstrations of most of the many illusory percepts that helped provide insight into the research, allowing participants to replicate the experiments in their own brains during the tutorial so that they can come to their own conclusions.

 

The phenomenology of perception and cognitive science: the continuing relevance of Merleau-Ponty's approach to conscious experience and thought

By Erik Myin (emyin@vub.ac.be)
Centre for Logic and Philosophy of Science, Department of Philosophy
Free University Brussels (VUB)

This workshop has a double aim :

  • to present some of the main threads of thinking of Merleau-Ponty's Phenomenology of Perception in terms of a vocabulary and framework that should be familiar to cognitive scientists
  • to explore the relevancy of the approach taken by Merleau-Ponty to various themes in the cognitive sciences, such as visual perception, synaesthesia and theory of mind.

The starting point will be Merleau-Ponty's idea that all conscious perception is a kind of action, grounded in an implicit acknowledgment of potential for further action. Besides the philosophical motives for this point of view, the neuro-scientific and psychological evidence referred to by Merleau-Ponty (lesion studies, distorting viewing conditions, phantom limbs, anosognosia, ?) will be considered and related to more recent work. Further analysis of 'experience as acknowledgement of potential for action' then leads to the role the whole body plays as the unit of perception and action. It is shown how unraveling the apparently strange claim that 'synaesthetic perception is the rule' leads to a theory of normal perception in which various notorious difficulties such as the 'qualitative nature' of sensation or the 'binding problem' loose their aura of intractability. As a bonus, it is argued, this account of normal perception can be naturally extended into a plausible theory of proper synaesthesia. Nowadays forgotten evidence from Merleau-Ponty's days is laid aside recent work. Then it is shown how Merleau-Ponty construes knowledge of other minds as involving a perceptual capacity. Again, this thesis and the evidence indicated by Merleau-Ponty, is related to recent experiments and theory. In conclusion, the pivotal distinction made by Merleau-Ponty between the 'subjective' and the 'objective' is investigated. It is argued that a refreshing view on the 'hard problem' of explaining consciousness can be derived from it, in which the latter arises out of an impossibility to take simultaneously two different views on a single reality, rather than on an incompatibility of different entities. It will be argued that some of the pessimism expressed by Merleau-Ponty on the convergability of the two points of view, might be no longer tenable given more recent views on causality. This is illustrated by reconsidering Merleau-Ponty's analysis of phantom limbs from within a 'dynamical systems' framework.

A comprehensive syllabus will be available to the participants.

 

Unconscious perception and its implications for consciousness

By Michael Snodgrass (jmsnodgr@umich.edu)

Historically, unconscious perception research has been concerned with whether stimuli that are not phenomenally conscious nonetheless produce effects. To answer this question requires that the conscious vs. unconscious distinction be clarified and that methods be developed that can disambiguate conscious vs. unconscious perceptual influences. Vigorous controversy continues over how to accomplish these goals (cf. Merikle & Reingold, 1998; Snodgrass, in press). For example, there is strong disagreement regarding whether subjective threshold (where participants deny awareness, but can nonetheless discriminate stimuli above chance) or more stringent objective threshold (where above-chance discrimination does not occur) indexes of consciousness are more appropriate. I suggest that close examination of these controversies has crucial implications not only for the status of unconscious perception research in particular but for theories of consciousness in general.

The workshop begins by reviewing the experimental and theoretical background, which has been characterized by longstanding methodological disagreement over the fundamental problem: How can we validly infer unconscious perceptual influences? In particular, how can alternative weak conscious perception explanations be ruled out? Classic dissociation paradigms approach this problem by seeking effects on a higher-level task intended to index unconscious perception (e.g., semantic priming), while simultaneously demonstrating null sensitivity on a lower-level task intended to index conscious perception (e.g., detection). As Reingold & Merikle (1988, 1990) made clear, however, it is difficult to specify which, if any, measures validly index conscious perception (the exhaustiveness problem), or how to insure that conscious perception has indeed been completely eliminated (the null sensitivity problem). Moreover, given that conscious perception indexes are plausibly sensitive to unconscious perceptual influences also, attaining null sensitivity may ironically eliminate them as well (the exclusiveness problem). These formidable problems, as well as the apparent unreliability of objective threshold effects, led to the development of the two currently dominant unconscious perception paradigms: 1) Merikle and associates' subjective threshold/qualitative differences model, and 2) Greenwald and associates' objective threshold/rapid decay model. These important paradigms are explored in detail, highlighting their many methodological and empirical contributions. I will argue, however, that these paradigms have not yet provided strong evidence for unconscious perception.

Snodgrass and associates' (in press; Snodgrass, 2002; Snodgrass, in press) methodological analysis is then presented, stressing a model-based approach. Demonstrating unconscious perception requires rejecting the conscious-perception-only model, which in turn requires specifying its key properties. Various evidence will suggest that conscious perception functions on a hierarchical strength/complexity continuum, such that greater stimulus intensity is required in order for more complex effects to occur. This central principle yields three conclusions: 1) Direct measures that index lower-level stimulus features (e.g., detection) are exhaustively sensitive with respect to effects that require higher-level processing (e.g., semantic priming); 2) Effects that correlate positively with suitable conscious perception indexes are consistent with the strength/complexity continuum, yielding weak evidence for unconscious perception; and 3) Effects which correlate negatively with such indexes violate the strength/complexity continuum, providing strong evidence for unconscious perception. Further, the exhaustiveness, null sensitivity, and exclusiveness problems predict positive relationships; hence, negative relationships overcome these difficulties. With this methodological framework in mind, the literature is reviewed and the objective threshold/non-monotonic model proposed. I argue that reliable effects occur at the objective detection threshold (ODT) but that null effects occur at the longer, objective identification threshold (OIT), suggesting a negative relationship in the ODT-OIT region and allowing strong inferences for unconscious perception at the ODT. I interpret this pattern as suggesting that conscious and unconscious perceptual influences are functionally exclusive, such that the former usually override the latter when both are present. The relationship is negative in the ODT-OIT region because emerging conscious perception is not yet sufficient to support higher-level effects (i.e., these stimuli are detectable but not identifiable). Conversely, conscious perception above the OIT (e.g., subjective threshold stimuli) is capable of supporting such effects; moreover, these effects are positively related with stimulus intensity, consistent with the conscious-perception-only model.

Implications of the objective threshold/non-monotonic model are then examined. A closer look at subjective threshold qualitative differences reveals that they are characterized by the presence vs. absence of metacognitive response strategies (e.g., whether stimuli are excluded or not). Under certain circumstances negative relationships emerge with subjective threshold stimuli as well, providing strong evidence for two conscious processes: Phenomenal consciousness per se and metacognitive awareness. The tripartite model of consciousness is then proposed: Stimuli can be either completely unconscious, phenomenally conscious but metacognitively unconscious, or both phenomenally and metacognitively conscious. Objective threshold approaches index phenomenal consciousness, whereas subjective threshold approaches index metacognitive consciousness. Importantly, these three processes are empirically separable and moreover exhibit qualitative differences. For example, responses to phenomenally unconscious stimuli cannot be consciously controlled, responses to metacognitively unconscious stimuli are potentially controllable but usually are not, and responses to fully conscious stimuli are straightforwardly controllable. Parallels with and implications for other models of consciousness are then explored, including Block's phenomenal vs. access consciousness, Baars' global workspace theory, and Rosenthal's higher-order thought (HOT) theory. Finally, practical guidelines for unconscious perception research are suggested, emphasizing the need for clarity as to whether phenomenal or metacognitive consciousness is being indexed by the particular stimulus intensities used.

As a final note, because signal detection theory plays a key role in interpreting unconscious perception research, participants in this workshop might also wish to take William Banks's "SDT and Consciousness Research" workshop offerred in the morning session.

Outline for the workshop:

I. First hour: Experimental and theoretical background

Early research and the signal detection theory critique

By Michael Snodgrass (jmsnodgr@umich.edu)

Historically, unconscious perception research has been concerned with whether stimuli that are not phenomenally conscious nonetheless produce effects. To answer this question requires that the conscious vs. unconscious distinction be clarified and that methods be developed that can disambiguate conscious vs. unconscious perceptual influences. Vigorous controversy continues over how to accomplish these goals (cf. Merikle & Reingold, 1998; Snodgrass, in press). For example, there is strong disagreement regarding whether subjective threshold (where participants deny awareness, but can nonetheless discriminate stimuli above chance) or more stringent objective threshold (where above-chance discrimination does not occur) indexes of consciousness are more appropriate. I suggest that close examination of these controversies has crucial implications not only for the status of unconscious perception research in particular but for theories of consciousness in general.

The workshop begins by reviewing the experimental and theoretical background, which has been characterized by longstanding methodological disagreement over the fundamental problem: How can we validly infer unconscious perceptual influences? In particular, how can alternative weak conscious perception explanations be ruled out? Classic dissociation paradigms approach this problem by seeking effects on a higher-level task intended to index unconscious perception (e.g., semantic priming), while simultaneously demonstrating null sensitivity on a lower-level task intended to index conscious perception (e.g., detection). As Reingold & Merikle (1988, 1990) made clear, however, it is difficult to specify which, if any, measures validly index conscious perception (the exhaustiveness problem), or how to insure that conscious perception has indeed been completely eliminated (the null sensitivity problem). Moreover, given that conscious perception indexes are plausibly sensitive to unconscious perceptual influences also, attaining null sensitivity may ironically eliminate them as well (the exclusiveness problem). These formidable problems, as well as the apparent unreliability of objective threshold effects, led to the development of the two currently dominant unconscious perception paradigms: 1) Merikle and associates' subjective threshold/qualitative differences model, and 2) Greenwald and associates' objective threshold/rapid decay model. These important paradigms are explored in detail, highlighting their many methodological and empirical contributions. I will argue, however, that these paradigms have not yet provided strong evidence for unconscious perception.

Snodgrass and associates' (in press; Snodgrass, 2002; Snodgrass, in press) methodological analysis is then presented, stressing a model-based approach. Demonstrating unconscious perception requires rejecting the conscious-perception-only model, which in turn requires specifying its key properties. Various evidence will suggest that conscious perception functions on a hierarchical strength/complexity continuum, such that greater stimulus intensity is required in order for more complex effects to occur. This central principle yields three conclusions: 1) Direct measures that index lower-level stimulus features (e.g., detection) are exhaustively sensitive with respect to effects that require higher-level processing (e.g., semantic priming); 2) Effects that correlate positively with suitable conscious perception indexes are consistent with the strength/complexity continuum, yielding weak evidence for unconscious perception; and 3) Effects which correlate negatively with such indexes violate the strength/complexity continuum, providing strong evidence for unconscious perception. Further, the exhaustiveness, null sensitivity, and exclusiveness problems predict positive relationships; hence, negative relationships overcome these difficulties. With this methodological framework in mind, the literature is reviewed and the objective threshold/non-monotonic model proposed. I argue that reliable effects occur at the objective detection threshold (ODT) but that null effects occur at the longer, objective identification threshold (OIT), suggesting a negative relationship in the ODT-OIT region and allowing strong inferences for unconscious perception at the ODT. I interpret this pattern as suggesting that conscious and unconscious perceptual influences are functionally exclusive, such that the former usually override the latter when both are present. The relationship is negative in the ODT-OIT region because emerging conscious perception is not yet sufficient to support higher-level effects (i.e., these stimuli are detectable but not identifiable). Conversely, conscious perception above the OIT (e.g., subjective threshold stimuli) is capable of supporting such effects; moreover, these effects are positively related with stimulus intensity, consistent with the conscious-perception-only model.

Implications of the objective threshold/non-monotonic model are then examined. A closer look at subjective threshold qualitative differences reveals that they are characterized by the presence vs. absence of metacognitive response strategies (e.g., whether stimuli are excluded or not). Under certain circumstances negative relationships emerge with subjective threshold stimuli as well, providing strong evidence for two conscious processes: Phenomenal consciousness per se and metacognitive awareness. The tripartite model of consciousness is then proposed: Stimuli can be either completely unconscious, phenomenally conscious but metacognitively unconscious, or both phenomenally and metacognitively conscious. Objective threshold approaches index phenomenal consciousness, whereas subjective threshold approaches index metacognitive consciousness. Importantly, these three processes are empirically separable and moreover exhibit qualitative differences. For example, responses to phenomenally unconscious stimuli cannot be consciously controlled, responses to metacognitively unconscious stimuli are potentially controllable but usually are not, and responses to fully conscious stimuli are straightforwardly controllable. Parallels with and implications for other models of consciousness are then explored, including Block's phenomenal vs. access consciousness, Baars' global workspace theory, and Rosenthal's higher-order thought (HOT) theory. Finally, practical guidelines for unconscious perception research are suggested, emphasizing the need for clarity as to whether phenomenal or metacognitive consciousness is being indexed by the particular stimulus intensities used.

As a final note, because signal detection theory plays a key role in interpreting unconscious perception research, participants in this workshop might also wish to take William Banks's "SDT and Consciousness Research" workshop offerred in the morning session.

Outline for the workshop:

I. First hour: Experimental and theoretical background

  • Early research and the signal detection theory critique
  • Subjective and objective indexes of consciousness
  • The Reingold & Merikle critique
  •  Merikle et al.'s subjective threshold model
  •  Greenwald et al.'s objective threshold/rapid decay model


II. Second hour: A model-based approach to method and evidence

  • Which measures validly index conscious perception?
  • Specifying the null hypothesis: Properties of the conscious-perception-only model
  • What kind of evidence warrants rejection of the conscious-perception-only model? That is, when are inferences for a second, unconscious perceptual process justified?
  •  The objective threshold/non-monotonic model

The functional exclusiveness postulate: conscious perceptual influences typically override unconscious ones

III. Third hour: Implications of the objective threshold/non-monotonic model:

  • The tripartite consciousness model: Perception can be 1) Completely unconscious, 2) Phenomenally conscious but metacognitively unconscious, or 3) Phenomenally and metacognitively conscious
  • Objective methods index phenomenal consciousness, whereas subjective methods index metacognitive consciousness
  • Qualitative differences between phenomenally unconscious and metacognitively unconscious phenomena
  • Implications for major models of consciousness, including Block's phenomenal vs. access consciousness
  • Baars' global workspace model
  • Rosenthal's higher-order thought (HOT) model
  • Towards a typology of neuropsychological disorders
  • Practical guidelines for unconscious perception research

Sources:

Snodgrass, M. (in press). Unconscious perception and its implications for consciousness. Philadelphia PA: John Benjamins.
Snodgrass, M., Bernat, E., & Shevrin, H. (in press). Unconscious perception: A model-based approach to method and evidence. Perception and Psychophysics. To appear with peer commentary.
Snodgrass, M. (2002). Disambiguating conscious and unconscious influences: Do exclusion paradigms demonstrate unconscious perception? American Journal of Psychology, 115, 545-580.

 

Chimpanzee minds

By Daniel Povinelli (ceg@louisiana.edu)

What theoretical approach is likely to lead to an accurate understanding of the minds of our closest living relatives, chimpanzees? One long-standing approach is to assume that the chimpanzee mind is simply a duller, less talkative version of the human mind; put the other way around, the human mind is seen as a brighter, more verbal version of the basic mammalian mind. This notion permeates current and historical; approaches to studying the mental faculties of chimpanzees and other nonhuman primates. While it is true that the human and chimpanzee mind may differ only parametrically with respect to certain cognitive systems, it is also true that their minds may differ qualitatively as well. This tutorial examines this possibility with respect to certain aspects of social and physical cognition, such as theory of mind and an understanding of causality. Current research is examined and evaluated and suggestions for future lines of research are offered.