New User Registration

In order to register on this site, you must first submit the passphrase below.

2000 - July 20/22 University of Cambridge Talk Abstracts - Page 2
Article Index
2000 - July 20/22 University of Cambridge Talk Abstracts
Page 2
All Pages

Session A - Friday 21 July 4-5.20 pm

An associative learning model of the illusory correlation effect

Robin A Murphy and Stefanie Schmeer

University of Hertfordshire Hatfield

Social psychologists have offered several hypotheses to explain the formation of stereotypes. One perspective is that they are the consequence of a perceived illusory correlation between group membership and certain behaviours. Hamilton and Gifford (1976) proposed that the infrequency of minority groups make them distinctive, thereby distorting people's perception of the relationship between group membership and behaviour. A modification to the Rescorla-Wagner (1972) model can also predict this effect by simply assuming that stimuli 1) acquire associations when presented together 2) associations representing stimuli that are not present on a given trial are also updated (Van-Hamme & Wasserman, 1994). One of its unique predictions is that the distorted perception should emerge preasymptotically before eventually diminishing. We report an experiment in which subjects were presented with vignettes describing members of either Group A or Group B performing either 'good' or 'bad' behaviours. The ratio of good/bad behaviours was always equivalent for members of both groups, but subjects experienced twice as many instances of group A members. The strength of the illusory correlation between group membership and good behaviour was assessed after either 18, 36, 54, or 72 trials. Results supported the predictions of the modified Rescorla-Wagner model.

Hamilton, D.L. & Gifford, R.K. (1976). Illusory correlation in interpersonal perception: A cognitive basis of stereotypic judgements. Journal of Experimental Social Psychology, 48, 5-17.

Rescorla, R.A. & Wagner, A.R. (1972). A theory of Pavlovian conditioning: Variations in the effectiveness of reinforcement and nonreinforcement. In A.H.Black & W.F. Prokasy (Eds.), Classical conditioning II: Current theory and research (pp. 64-99). New York: Appleton-Century-Crofts.

Van-Hamme, L.J. & Wasserman, E.A. (1994). Cue competition and causality judgment: The role of the nonpresentation of compound stimulus elements. Learning and Motivation, 25, 127-151.

Associatively mediated anti-generalisation in memory

I P L McLaren

University of Cambridge

Consider two stimuli, A and B, whose representations are associatively linked such that presentation of one stimulus retrieves a representation of the other. If we now present A what happens to the representation of B after retrieval? A straightforward analysis might have it that whatever happens to the representation for A (perhaps association with some outcome) will also happen to B, though perhaps to a lesser extent. This would constitute a form of generalisation between A and B mediated by the associations between them. Another analysis, suggested by recent research on retrospective revaluation effects in humans and infra-humans (e.g. Dickinson and Burke, 1996), would predict that whatever happened to the representation of A the opposite might be expected for the representation of B. Thus, if an association between the representation for A and some outcome were strengthened, then it should be weakened in the case of B. We might term this associatively mediated anti-generalisation.

The research reported here focuses on the latter phenomenon, but in memory rather than learning. Using a same / different recognition task with human subjects, I first demonstrate that, under certain circumstances, the effect of associatively retrieving a stimulus representation is the opposite to that which occurs on presentation of the stimulus. Then I consider to what extent this effect aligns with the retrospective revaluation literature referred to earlier.

Dickinson, A. & Burke, J. (1996). Within-compound associations mediate the retrospective revaluation of causality judgements. Quarterly Journal of Experimental Psychology, 49B, 60-80.

Two new forms of retrospective revaluation in human contingency learning.

Jan De Houwer (Introduced by Dr David Shanks)

University of Southampton

When a compound cue AT is followed by an outcome (AT+), human participants will judge the relation between cue T and the outcome to be less strong if A alone was previously paired with the outcome (A+). According to the probabilistic contrast model, such blocking effect is due to the fact that the A+ trials allow one to infer that A on its own is sufficient to explain the occurrence of the outcome on the AT+ trials. However, this conclusion is only justified if (a) A+ trials are treated as trials on which A and the outcome are present but T is absent and if (b) A is not accompanied by another predictive cue on the A+ trials. I report the results of two experiments which support the hypothesis that retrospective inferences about the status of T and other alternative cues on the A+ trials influence the contingency judgements for cue T. These findings support the probabilistic contrast model but are incompatible with the (revised) Rescorla-Wagner and SOP models.

Normative and descriptive accounts of the influence of power and contingency on causal judgment

David R Shanks1 and J Perales2

1. University College London

2. University of Granada, Spain

The Power PC theory (Cheng, 1997) predicts that causal judgments are based on causal power p of a potential cause, where p is the cause-effect contingency normalized by the base rate of the effect. Previous reports have demonstrated that both cause-effect contingency and effect base-rate independently affect estimates in causal learning tasks. In the present work these effects are replicated in three experiments. In Experiment 1, causal strength judgments were directly related to power p in a task in which the effect base-rate was manipulated across two positive and two negative contingency conditions. However, in Experiments 2 and 3 contingency manipulations affected causal estimates in several situations in which power p was held constant, contrary to the Power PC theory's predictions. This effect cannot be explained by participants' conflation of reliability and causal strength, as Experiment 3 demonstrated independence of causal judgments and confidence. Neither the Power PC theory nor the Rescorla-Wagner model can account for the entire pattern of results. However, the data are compatible with Pearce's (1987) model, as well as with the Belief Adjustment model (Catena, Maldonado, & Cßndido, 1998).

Session B - Friday 21 July 4-5.20 pm

Motor activation and inhibition elicited by masked primes: A threshold model and experimental evidence

Friederike Schlaghecken and Martin Eimer

Birkbeck College, London

Inhibitory processes play an important role in information processing and motor control. They are usually observed under conditions of supra-threshold stimulation. Near-threshold stimuli generally produce facilitation rather than inhibition. Recently, however, evidence for 'near-threshold inhibition' has been obtained (Eimer & Schlaghecken, 1998; Schlaghecken & Eimer, in press): Visual target stimuli were preceded by masked primes mapped to the same response as the target ('compatible'), or to the opposite response ('incompatible'). When primes were presented at fixation, responses in compatible trials were delayed, presumably reflecting an inhibition of the motor response corresponding to the prime. In contrast, with prime stimuli presented peripherally, responses were faster in compatible trials. We present evidence that this "central/peripheral asymmetry" reflects a threshold mechanism in motor control. Strong partial response activations (as elicited by foveal primes) are actively inhibited once the masking procedure removes the sensory evidence for the respective response tendency. In contrast, the weaker activations elicited by peripheral primes remain below the 'inhibition threshold,' and thus do not receive active inhibition. The finding that lowering the perceptual quality of central primes and enhancing the perceptual quality of peripheral primes reverses the central/peripheral asymmetry supports this idea.

Eimer, M. & Schlaghecken, F. (1998). Effects of masked stimuli on motor activation: Behavioral and electrophysiological evidence. Journal of Experimental Psychology: Human Perception and Performance, 24, 1737-1747.

Schlaghecken, F. & Eimer, M. (in press). A central/peripheral asymmetry in masked priming. Perception & Psychophysics.

Selection-for-action: Evidence from bimanual reach-to-grasp movements directed toward single and multiple 'objects'

Georgina M Jackson and Stephen R Jackson

University of Nottingham

According to the integrated competition hypothesis (Duncan, Humphreys, and Ward, 1997) visual information from a single object requires less processing resources than visual information from two separate objects. Here we report four experiments which investigated this distinction in relation to planning and control of visually guided, bimanual, reach-to-grasp movements. We found that different patterns of interference were observed dependent upon whether two target objects were perceived as separate objects, or were unified, either perceptually or functionally, to form a single object. We observed that when participants executed bimanual reach-to-grasp movements toward two differently-sized cylindrical dowels (incongruent reaches) that were not connected to one another, the maximum grip apertures for each hand were significantly larger than when reaching for two unconnected dowels of the same size (congruent reaches). In contrast, when participants executed bimanual reach-to-grasp movements toward the same dowels which were now unified, either perceptually or functionally, a different pattern of effects on grip aperture were observed. In the latter case, during incongruent reaches, the maximum grip apertures of each hand became more similar (i.e. maximum grip aperture to the large target item was reduced while that to the small target item increased). These results suggest that the processing demands required by bimanual movements can vary as a function of the perceived unity of the target items.

Duncan J, Humphreys GW, and Ward R. (1997) Competitive brain activity in visual attention. Current Opinion in Neurobiology, 7, 255-261.

Understanding projectile acceleration

Marco Bertamini1 and Heiko Hecht2

1. University of Liverpool

2. Man-Vehicle Lab, MIT, Cambridge, MA, USA

Throwing and catching balls or other objects is a generally highly practiced skill for young adults, however, conceptual as well as perceptual understanding of the mechanics that underlie this skill is surprisingly poor. In 5 experiments, we investigated conceptual and perceptual understanding of simple ballistic motion. Paper-and-pencil tests revealed that up to half of all participants mistakenly believed that a ball will continue to accelerate after it has left the thrower's hand. Observers also showed a remarkable tolerance for anomalous trajectory shapes. Perceptual judgments based on graphics animations replicated these erroneous beliefs for shallow release angles. Observers' tolerance for anomalies tended to decrease with their distance from the actor.

These findings are not consistent with claims in the naïve physics literature that liken intuitive beliefs to Aristotelian or medieval physics theories. Instead, observers seem to project their intentions to the ball itself (externalization) or even feel that some power is still exerted on the ball when still close.

Implicit knowledge and motor skills: What people who can catch a ball do not know

Peter McLeod1, Nick Reed1 and Zoltan Dienes2

1. Oxford University

2. Sussex University

Ball catching, like many motor skills, is under the control of implicit knowledge. That is, people are unable to describe how they choose an appropriate speed to run at to intercept a ball before it hits the ground. People are not only unable to describe their interception strategy, they cannot describe the sensory information which controls this strategy. Given a description of this information, recognition is better than free recall. But even if the recognition judgement is made just after catching a ball, and the subjects have been primed that they will be asked to make the judgement, it is still far from perfect. The experiments described will give those members of the EPS who are in denial about the existence of implicit knowledge the grounds to make a graceful exit from this implausible state.

Session C - Friday 21 July 4-5.20 pm

The role of androgens and estrogens in the Bruce effect

Denys deCatanzaro

McMaster University

Novel males can disrupt early pregnancy in previously inseminated female mice. Castrated males lose this ability. Through ELISA procedures, we have measured levels of testosterone and 17-beta estradiol in the excretions of intact and castrated novel males while they are housed near pregnant females. Significant levels of both of these steroids are found in the urine and feces of intact males. Novel intact males become exceptionally agitated in their behaviour while near inseminated females, and actively direct their urine at these females. When 17-beta estradiol is painted directly on the nasal area of inseminated females, pregnancy reliably fails. Apart from male urine, no other chemical, including testosterone, produces such a pregnancy loss. It is established that excessive levels of 17-beta estradiol can act at the uterus and fallopian tubes to disrupt intrauterine implantation of fertilized ova. Futhermore, actions of 17-beta estradiol at the hypothalamus can interfere with pregnancy by stimulating estrus.

Effects of the benzodiazepine receptor agonist chlordiazepoxide & the 5-HT1A antagonist WAY100635 on behaviour in the rat elevated zero maze are not influenced by the light/dark cycle

Lianne Stanford1, Scott M Weiss2 and Kelly J Stanhope2

1. Department of Psychology and Neuroscience, Dalhousie University, Halifax

2. Department of Neuropharmacology, Cerebrus Ltd., Wokingham

The rat elevated zero maze is a validated model of anxiety (Shepherd et al. 1994 Psychopharm 116:56-64). In this test, rats administered anxiolytic drugs typically show a behavioural profile of increased exploratory activity (line crosses & head dips), increased time in the open areas, and less risk assessment (stretch-attends). In mice, 5-HT1A receptor antagonists have altered profiles on the elevated plus-maze when tested during different portions of the light/dark cycle (Rodgers et al., 1998 J Psychopharm 12:A32). To investigate this further, rats housed under normal or reversed light/dark cycles were tested with chlordiazepoxide (CDP), a benzodiazepine receptor agonist or with WAY100635, a 5-HT1A receptor antagonist. CDP (4 mg/kg) treated rats showed a similar anxiolytic profile in both housing groups. WAY100635 (0.01 to 0.3 mg/kg, sc: 30min) was without effect on rats from either housing group. These data show that the time of testing relative to the diurnal cycle does not alter the profiles of either CDP or WAY100635, and question the assumption that time of testing is a critical determinant of the anxiolytic properties of 5-HT1A receptor antagonists.

The effect of long-term amygdala kindling on emotional behavior may be related to receptor regulation within the dentate gyrus.

Lisa E Kalynchuk1, Janet Menard2 and Michael J Meaney2

1. Dalhousie University

2. Douglas Hospital Research Center, McGill University.

Long-term amygdala kindling in rats produces dramatic increases in emotional behavior. For example, rats subjected to 100 amygdala stimulations display extreme resistance to capture, escape behavior in an elevated plus maze, and active defensive behavior in a resident-intruder paradigm (Kalynchuk et al., 1997; Kalynchuk et al., 1999). In this paper, we present evidence that kindled rats also display heightened acoustic startle, altered patterns of open-field exploration, and significant weight gain. We also summarize the results of our recent studies investigating changes in receptor expression that may underlie the development of kindling-induced emotionality. Remarkably, most of the receptor subtypes we have investigated (i.e., 5-HT, GABA/BZ, glucocorticoid, NMDA) are changed specifically in the dentate gyrus. The dentate gyrus is an important mediator of seizure activity, contextual memory, and stress sensitivity. Because kindling-induced emotionality is triggered by exposure to novel situations, we propose that abnormal functioning within the dentate gyrus may interfere with the kindled rats' ability to cope with stress (e.g., novelty), thus leading to overly-emotional responses. (Supported by research grants from NSERC and MRC).

Session A – Saturday 22 July 9-10.40

Objects look larger with the left eye than the right eye

Chris McManus and Julia Tomlinson

University College London

Very little research has compared perception in the right and left eyes, although much has compared the right and left visual fields of the two eyes. An intriguing exception (S. Coren and C. Porac. Nature 260:257-258, 1976) suggested that objects appeared larger in the right eye of right-eyed individuals and in the left eye of left-eyed individuals. The present study tried to replicate that result using a tachistoscope which allowed presentation of images separately to right and left eyes. Each stimulus consisted of a pair of solid circles which appeared in two of four locations diagonally above or below and left or right of the fixation point. Subjects’ task was to report which circle appeared larger. The design therefore allowed eye effects to be distinguished from visual field effects. 43 subjects were tested, 19 of whom were left-eyed and 10 of whom were left-handed. Overall, circles appeared about 1.6% greater in diameter in the left eye (3.2% difference in area). There was no effect of visual field, suggesting the effects were not due to conventional cerebral lateralisation. No evidence was found for Coren and Porac’s correlation with sighting eye dominance. There were however very significant individual differences which correlated most strongly with writing hand preference (rather than throwing hand). The mechanism of these eye differences raises interesting questions for theories of lateralisation and for theories of binocular vision, where, as a part of ‘eye signature’, they may be important for disambiguating input from the two eyes.

The medial forebrain bundle must interact with the cortex for normal object recognition memory

Alex Easton (Introduced by Dr D Gaffan)

University of Oxford

In monkeys, lesions of the temporal stem, amygdala and fornix produce a severe impairment in recognition memory performance. To test the hypothesis that this impairment was due to isolating the inferior temporal cortex from the basal forebrain, we tested animals with asymmetric lesions of the medial forebrain bundle in one hemisphere (to disrupt basal forebrain activity) and of the inferior temporal cortex in the opposite hemisphere. A similar level of impairment was seen in these animals in object recognition memory as in animals with sections of the temporal stem, amygdala and fornix. This supports the hypothesis that the basal forebrain is essential for normal memory in the primate. A similar effect from asymmetric lesions of the medial forebrain bundle and the frontal cortex suggests that the basal forebrain acts as a route of subcortical communication between the frontal and inferior temporal cortices in learning.

Response reversal impairment in psychopathic individuals

R J R Blair

Institute of Cognitive Neuroscience, University College London

Psychopathy is a severe developmental disorder involving an emotional impairment, characterized by a lack of guilt and empathy, and extremely antisocial behaviour (Hare, 1991). While many of the impairments of psychopathic individuals appear to be explainable in terms of amygdala dysfunction, there is also a suggestion of dysfunction in some of the functions mediated by orbitofrontal cortex. This paper reports on the performance of psychopathic individuals on two measures of orbitofrontal cortex functioning: the ID-ED task (Dias et al., 1996) and the four pack card playing task (Bechara et al., 1997). 14 psychopaths and 14 age and IQ matched, incarcerated controls, identified using the Revised Psychopathy Checklist (Hare, 1991), were presented with the two measures. The psychopathic individuals showed striking selective deficits. On the the ID-ED task, the psychopathic individuals showed a selective deficit in response reversal but no impairment in attentional set shifting. On the four pack card playing task, the psychopathic individuals were as likely as the controls to shift from the high reward but net loss packs to the low reward but net gain packs. However, the psychopathic individuals were significantly less likely than the controls to shift their pack choice following punishment (point loss). The results are interpreted with reference to dysfunction within the connections between the amygdala and orbitofrontal cortex.

A combined neuropsychological and structural MRI study of high functioning autism

Jill Boucher1, Patty Cowell2, Paul Broks2, Neil Roberts3, Matthew Howard3 and Andrew Mayes2

1. University of Warwick

2. University of Sheffield

3. University of Liverpool

A study was carried out assessing the hypotheses that autism is associated with pathology of the medial temporal lobes (MTL) and/or the prefrontal cortex (PFC) (Robbins, 1997), using a combination of neuropsychological tests and sMRI. The sMRI investigations are described more fully in a related paper/poster.

Ten males with high functioning autism (HFA) aged 16 – 40 years, with verbal IQ’s between 86 – 150, and 10 normal controls individually matched for sex, age, and verbal IQ were tested. The HFA diagnoses were confirmed using a diagnostic checklist (Wing, 1996). The neuropsychological tests assessed executive functions (Zoo map, Brixton, Haylings, FAS), social cognition (Benton, Ekman, Eye Direction Detection, Point-Light Movement Interpretation, Face Recognition), and memory (Face and Word Recognition, Picture Recognition, Story Recall, Figure Copying and Recall), these tests being sensitive to pathology affecting the PFC, the amygdala, and hippocampal-related regions, respectively.

There were significant between group differences on the Zoo Map and Haylings; the Ekman (especially fear), Face Recognition, Eye Direction and Point Light tests. The MRI scans showed autism-specific effects in the MTL, especially the amygdala; and PFC asymmetry similar to that which occurs in other neurodevelopmental disorders. These findings support a MTL (amygdala) hypothesis, and give qualified support to a PFC hypothesis.

A selective impairment in the recognition of facial and vocal signals of disgust following brain injury.

Andrew J Calder1 , Jill Keane1 and Andy Young2

1. MRC Cognition and Brain Sciences Unit, Cambridge

2. University of York

Patient studies and brain-imaging work have shown that the recognition of facial signals of disgust may be linked to a distinct neural substrate; two areas that have been suggested are the fronto-striatal regions and the insula. (Phillips et al., 1997; Sprengelmeyer et al., 1996). Central to our interpretation of these studies is whether they have identified a brain system that is involved specifically in the perception facial signals of disgust, or one that underlies the perception disgust signals from all sensory modalities. We address this issue in a case study of NK, a 25 year old male with damage to left basal ganglia and sub-insula.

NK's ability to identify facial expressions was examined using two tests. The first contained ten examples of each of six facial expressions (happy, sad, anger, fear, disgust, and surprise) from the Ekman and Friesen (1976) series. The second included eight examples of each of seven facial expressions (the six listed above plus contempt) from the Matsumoto and Ekman (1988) set. On each test, NK showed a selective impairment in recognising facial expressions of disgust.

NK's recognition of emotion from vocal expressions was also assessed using two separate tests. The first used emotional vocalisations (e.g., laughter for happiness, retching for disgust, etc.) and contained ten examples of each of the six emotions listed above. The second comprised individual lists of five random digits, with each list spoken with intonations intended to convey particular emotional prosodies (happy, sad, anger, fear, or disgust). For both vocal tests, NK showed a significant impairment for recognising vocal signals of disgust. With the exception of a mild deficit for surprise on the first test, none of the other emotions was impaired significantly.

NK's results are consistent with damage to a brain system geared to interpret signals of disgust regardless of their modality.

Phillips, M. L., Young, A. W., Senior, C., Brammer, M., Andrew, C., Calder, A. J., Bullmore, E. T., Perrett, D. I., Rowland, D., Williams, S. C. R., Gray, J. A., & David, A. S. (1997). A specific neural substrate for perceiving facial expression of disgust. Nature, 389 (2 October), 495-498.

Sprengelmeyer, R., Young, A. W., Calder, A. J., Karnat, A., Herwig, L., Homberg, V., Perrett, D. I., & Rowland, D. (1996). Loss of disgust in Huntington's disease: Perception of faces and emotions. Brain, 119, 1647-1665.

Session B - Saturday 22 July 9-10.40

Performance as a function of number and spatial arrangement of food locations on a large radial maze

Mark R Cole

Huron College, University of Western Ontario

Cole (1999) showed that when 8, 16, 24, 32, 40 and 48 baited arms were added successively in random locations to a large radial maze, performance by rats declined only when the final set of eight arms was added. The present research was designed to assess whether this decline was primarily due to number of food locations or to configuration of food locations. In successive phases, each lasting 20 sessions, 8, 16, 24, or 48 baited arms were added to a large radial maze. Rats were tested twice daily on the maze. In one daily trial, the arms were spread out as far apart as possible, while in the other daily trial, the arms were located as tightly together as possible. Location of arms proved to be more important than number of arms when the percentage of novel entries in the first N opportunities was examined. In the Spread Condition, performance was nearly perfect with only eight arms attached, but declined as the number of arms increased to 16, 24 and 48. In the Tight Condition, performance was poor even with only eight arms attached, and declined only marginally as further arms were added. The results were interpreted as demonstrating that performance on a radial maze is more dependent upon the angular separation between adjacent arms than on the number of arms.

Symposium: "Spatial cognition in animals"

Organiser: Ron Weisman

Visual Scene Analysis in Pigeons

Marcia Spetch

University of Alberta

The world consists of a complex array of global and local visual cues that might be used to identify and remember a particular location, such as a place where food is found. Pigeons can find hidden goals using any of a variety of spatial cues, including single landmarks, arrays of landmarks, distance from an edge, and geometric shape of the environment. Pigeons' use of spatial information is remarkably general across tasks conducted on computer screens or the laboratory floor, despite drastic differences in scale of space and type response. Comparative studies have revealed interesting similarities and differences between how pigeons and other species use of spatial information.

Seasonality and spatial memory in birds

Sue Healy and Robert Biegler

University of Edinburgh

Seasonal variation in song is related to volume changes in brain regions associated with song learning. Recent evidence suggests that hippocampal volumes also vary through the year: they are large in autumn/winter and small in spring/summer. This variation appears to correlate with changes in food-storing behaviour (and thus memory for locations of stored food). We tested food-storing and non-storing tits on spatial and nonspatial memory tasks using birds caught in autumn and spring. We found seasonal differences in memory performance but, counter to prediction, in both species. It is not clear, whether there are correlated neuroanatomical changes.

Place finding in insects: Stereotypical servomechanisms

Ken Cheng

Macquarie University, Australia

Insects use a series of stereotypical place-finding servomechanisms to return to a place. Use of path integration gives way to use of landmarks close to the target location. Sun-compass and large landmarks provide directional cues. Visual flow provides an estimate of distance traveled. Landmark-based servomechanisms include beaconing: heading towards a recognized landmark, based on multiple stored views of the landmark, sensory-motor vectors: motor trajectories triggered by particular (sensory) views of a landmark, and image matching: positioning landmarks at correct retinal coordinates while facing a stereotypical direction. A 'simple-minded' sequence of servomechanisms, while not 'map-like', does the job.

Auditory Distance Perception in Black-Capped Chickadees

Leslie Phillmore, Chris Sturdy and Ron Weisman

Queen's University

In the wild, songbirds can use auditory cues to determine spatial distance to conspecifics and to identify individual conspecifics. We have been conducting perceptual experiments in the laboratory to study the auditory cues black-capped chickadees use to determine distance. Chickadees can use the amplitude of a vocalization to estimate the distance to the signal. Chickadees reared in isolation from conspecifics are not deficient in the use of auditory cues to judge distance to a vocal signal. It appears that the ability to use auditory spatial cues is not appreciably effected by experience with conspecific vocalizations at a distance.

(End of symposium)

Session C - Saturday 22 July 9-10.40

A Practical Metric for the Complexity of Graphic Displays

D C Donderi

McGill University, Montreal

A practical and effective complexity measure for character-based displays was developed by Tullis in 1983, and a comparable measure for graphical displays would be very useful. A possible measure of graphical image complexity is the change in file size after a bitmap image has been compressed. The most informationally redundant image can be compressed into the smallest file, while the least informationally redundant image can be compressed the least. Magnitude estimation scaling (MES) ratings of the subjective complexity of three different sets of 50 graphics bitmap images were made by three different groups of 10 observers each. For all three groups, the file size resulting from from JPEG image compression was highly correlated with MES judgments of subjective complexity of the original bitmap files. The relationships were r = 0.73 for a set of 50 electronic radar displays, r = 0.65 for a set of 50 electronic chart displays, and r = 0.57 for a set of 50 overlay displays containing both radar and electronic chart imagery. Work is in progress to investigate the properties of other compression algorithms.

Mechanisms for detecting texture gradients

Frederick A A Kingdom and Anthony Hayes

McGill Vision Research Unit, Montréal

Texture gradients arise in the retinal-image projection of any non-coplanar textured surface, and provide important monocular cues to surface shape. A widely held view is that texture gradients are detected by a cortical mechanism that can be approximated by a Filter-Rectify-Filter (FRF) model, where local features are picked up by Œ1st stage¹ filters (such as simple-cells) and, after conversion to energy responses via rectification (or squaring), are pooled by Œ2nd-stage¹ filters that detect variations in response energy across space. Because an FRF mechanism deals only with filter-response energy, it is sensitive to a variety of texture gradients (e.g., contrast, size, orientation), and subthreshold amounts of different texture gradients can be predicted to summate to detection threshold. We tested this prediction with stimuli that consisted of dense arrays of Gabor elements (windowed sinusoids), whose orientations, spatial frequencies, and contrasts, were modulated sinusoidally across space. Using a conventional test-threshold-versus-mask-amplitude paradigm, we found no instances of subthreshold summation between different texture gradients. We conclude that rather than using FRF mechanisms to indiscriminately detect a variety of different texture gradients, the visual system employs separate mechanisms for detecting different types of texture gradient.

Motion of drifting "isoluminant" chromatic gratings is mediated by a luminance mechanism.

Kathy T Mullen, Tatsuya Yoshizawa and Curtis L Baker.

McGill Vision Research, Department of Ophthalmology, McGill University

In previous work using random gabor kinematograms we found that there was no linear motion for red-green color vision (Yoshizawa, Mullen & Baker, Vis. Res, in press). Depending on the conditions, the motion of these isoluminant stimuli was supported either by a luminance "artifact" or by a chromatic nonlinear motion mechanism. We now ask what type of mechanism supports the motion of drifting chromatic gratings.

We measure the masking of drifting isoluminant red-green gratings (1.5cpd) by dynamic broad spectrum luminance and chromatic noise. Our results show that luminance noise masks direction discrimination thresholds of chromatic gratings but not detection thresholds over a wide range of drift rates (0.75-9Hz). On the other hand, chromatic noise masks chromatic detection thresholds, but has much less effect on direction discrimination, and at high drift rates chromatic noise has very little effect on either chromatic threshold.

The double dissociation of the effects of color and luminance noise masking suggests that the motion of drifting "isoluminant" chromatic gratings is mediated solely by a luminance mechanism.

A colour concept in male marmoset monkeys

Andrew M Derrington, Amanda Parker and Greg G Goodson

University of Nottingham

The colour vision of the male marmoset monkey (Callithrix Jacchus) is a primitive dichromatic (two-receptor) system[1]. Dichromatic colour vision may not support an innate concept of colour: it discriminates between monochromatic colours but confuses white light with monochromatic light that produces the same ratio of excitations in the two receptor types[2]. Here we show that male marmosets learn rapidly to distinguish between pairs of abstract patterns that differ only in that one member of each pair is grey-scale and the other coloured. The discrimination is not affected by manipulating the lightness of the grey-scale pattern elements and transfers to new pairs of patterns that contain only colours not included in the training patterns. We conclude that dichromatic colour vision supports an innate concept of colour and that marmosets are aware of using colour to solve perceptual problems. This may be the first demonstration that a non-human species has a concept of colour.

1. Travis, D.S., J.K. Bowmaker, and J.D. Mollon, Polymorphism of visual pigments in a callitrichid monkey. Vision Research, 1988. 28(4): p. 481-490.

2. Cornsweet, T.N., Visual Perception. 1970, San Diego: Harcourt Brace Jovanovich.

Impoverished second-order input to global linking in human vision

Robert F Hess, Tim Ledgeway and Steven Dakin

McGill Vision Research, McGill University

Department of Visual Science, University of London

Recent evidence points to the importance of global operations across spatial regions larger than individual cortical receptive fields. Studies of contour integration and motion trajectory detection suggest that network operations between local detectors underlies the encoding of extended contours in space and extended trajectories in motion. Here we ask whether such network operations also occur between second-order-detectors known to exist in visual cortex. We compared performance for stimuli composed of either first-order or second-order elements equated for visibility, and we show that unlike the first-order case, there is little or no linking interaction between local second-order detectors. This implies that the network operations thought to underlie the integration of contour information across space and direction information over space and time receive, at best, an impoverished input from local detectors that encode second-order image attributes.

Session A – Saturday 22 July 11.20-1 pm

Symposium: "Neuropsychology of human long-term spatial memory"

Organisers: R G Morris and A R Mayes

Path integration following unilateral temporal lobectomy

R G Morris1, C Worsely1, H Spiers2, C E Polkey3 and M Recce4

1. Institute of Psychiatry

2. Institute of Cognitive Neuroscience, University College London

3. Academic Neurosurgery Unit, The King's Neuroscience Centre

4. New Jersey Institute of Technology, Newark, USA.

Path integration, a component of spatial navigation, is the process used to determine position information on the basis of information about distance and direction travelled derived from self-motion cues. Following on from studies in the animal literature that seem to support the role of the hippocampal formation in path integration, this facility was investigated in humans with focal brain lesions. Thirty-three neurosurgical patients (17 left temporal lobectomy, LTL; 16 right temporal lobectomy, RTL) and sixteen controls were tested on a number of blindfolded tasks designed to investigate path integration, and on a number of additional control tasks (assessing mental rotation and left-right orientation). In a test of the ability to compute a homing vector, the subjects had to return to the start after being led along a route consisting of two distances and one turn. Patients with RTL only were impaired at estimating the turn required to return to the start. On a second task, route reproduction was tested by requiring the subjects to reproduce a route consisting of two distances and one turn; the RTL group only were also impaired at reproducing the turn but this impairment did not correlate with the homing vector deficit. There were no group differences on tasks where subjects were required to reproduce a single distance or a single turn. The results suggest that path integration is impaired patients with RTL only and suggest that the right temporal lobe plays a role in idiothetic spatial memory.

The temporal lobes, navigation and memory in large-scale virtual space

H J Spiers1, N Burgess1, E A Maguire2, S A Baxendale3, T Hartley1 and J O'Keefe1

1. Institute of Cognitive Neuroscience, University College London

2. Wellcome Department of Cognitive Neurology, Institute of Neurology, University College London

3. National Hospital for Neurology and Neurosurgery, London

Unilateral damage to the right medial temporal region is known to cause impairments on a number of spatial memory tasks, whereas bilateral damage involving both hippocampi results in substantial episodic memory deficits. In the present study we used a large-scale virtual reality town to test topographical and episodic memory in patients with unilateral medial temporal lobe damage. Sixteen left and 16 right temporal lobectomy patients were compared with 16 healthy matched control subjects. Having explored the town at length, topographical memory was then assessed by requiring subjects to navigate in, recognise scenes from and draw maps of the virtual town. Following this, subjects collected objects from two different characters in two different locations within the town. Episodic memory of these events was then assessed by a paired forced choice recognition test. Right temporal lobectomy patients were impaired on navigation, scene recognition and computer aided map drawing relative to control subjects. They were also mildly impaired on recognition of objects in the episodic memory task. The left temporal lobectomy patients were impaired relative to control subjects on their ability to remember who gave them the objects and the order in which they received the objects in the episodic memory task. Both patient groups were at chance on their ability to remember where they received the objects. These results suggest that topographical memory is largely mediated by the right medial temporal lobe, whereas some associative aspects of episodic memory are more dependent on the left medial temporal lobe.

Spatial memory and the hippocampal region: A structural MRI analysis of patients with unilateral mesial temporal lobe sclerosis

S. Abrahams1, R. G. Morris1, C. E. Polkey2 and A. Pickering3

1. Institute of Psychiatry

2. Academic Neurosurgery Unit, The King's Neuroscience Centre

3. St. George's Hospital Medical School, London,

The Nine-Box Maze, (Abrahams et al. 1997) was designed to compare spatial mapping and working memory theories of the functions of the hippocampus. The task provides separate measures of spatial, object, working and reference memory, and is analogous to the radial arm maze, encouraging the formation of allocentric rather than egocentric spatial representations. In the current investigation 47 patients with unilateral temporal lobe epilepsy (TLE) were investigated. Spatial memory deficits across both working and reference memory conditions were found in patients with a right epileptogenic focus. There was no evidence of an object working memory deficit, but a non-lateralised impairment in object reference memory was revealed. These results extend our previous findings in a larger group of patients. Moreover, the pattern of results was confirmed in a subgroup of 33 patients with unilateral atrophy localised to the mesial temporal lobe structures as revealed by volumetric analysis of magnetic resonance images. The structural analysis revealed evidence of unilateral sclerosis in the hippocampus and parahippocampal gyrus, but which did not extend into the remaining temporal cortex. In addition, spatial memory errors significantly correlated with volumetric measures of the mesial temporal lobe structures. In contrast object reference memory errors correlated with volumetric measures of the temporal cortex and not with mesial temporal lobe structures. These findings support a specialised role for the right hippocampal and parahippocampal region in spatial memory.

Spatial memory dissociations following lesions to the hippocampus and other medial temporal lobe regions

J S Holdstock1, A R Mayes2, J P Aggleton3 and N Roberts4

1. University of Liverpool

2. Department of Clinical Neurology, University of Sheffield

3. Cardiff University

4. Magnetic Resonance and Image Analysis Research Centre, University of Liverpool

The spatial memory of a patient (YR) with relatively selective bilateral hippocampal damage was investigated. Location memory was tested under conditions which encouraged the use of either an allocentric (identification of a location by its position relative to an array of external landmarks) or egocentric (identification of a location by its position relative to the observer) frame of reference. The task required subjects to remember the position of a single light presented on a board. In the allocentric condition subjects moved to a different place around the board before making a mnemonic response. In the egocentric condition stimuli were presented in darkness, which eliminated allocentric cues. YR was found to be more impaired at recalling allocentric than egocentric information after a sixty second filled interval with a tendency for the impairment to increase up to this delay. Recognition of allocentric spatial information was also impaired at the sixty second delay. More extensive medial temporal lobe damage which included the parahippocampal, perirhinal and entorhinal cortices in addition to the hippocampus (patient NM) also resulted in a deficit in allocentric but not egocentric spatial recall and this was of equivalent severity to that seen in YR. The results suggest that the human hippocampus has greater involvement in allocentric than egocentric spatial memory and that this most likely concerns the consolidation rather than initial encoding of allocentric information into long-term memory. However, a further task showed that YR's recall and recognition of the positions of specific pictures arranged on a table top was impaired even though an egocentric frame of reference could have been used. This result leads us to consider whether the allocentric spatial memory deficit reflects a more general problem of storing or retrieving certain kinds of associative information.

Right medial-temporal contributions to human episodic memory for object location and object identity in visual scenes: Evidence from functional neuroimaging

S Köhler, B. Milner and J. Crane

Montreal Neurological Institute, McGill University, Canada

Previous neuroimaging research has implicated the hippocampus and the posterior parahippocampal gyrus in encoding novel visual scenes. In the present PET study, we investigated whether these structures are also involved in recognition of spatial or object information from previously encountered scenes. Healthy participants were scanned while encoding photographs of novel scenes and while performing three different forced-choice recognition memory tests. These tests required the discrimination of familiar from altered scenes, which differed from each other either in the identity of one of the objects, the spatial configuration the objects formed, or specific object-place associations within identical configurations. Participants were also scanned in two visual baseline tasks and during encoding of individual objects presented without a scene context. Encoding of novel scenes, as compared to baseline, activated anterior portions of the hippocampus and posterior aspects of the parahippocampal gyrus bilaterally. A comparison of scene encoding with encoding of individual objects showed that the parahippocampal activation extended into a right-sided region previously described as parahippocampal place area (PPA). Recognition of scenes, as compared to baseline, activated the PPA but not the hippocampus, irrespective of whether the recognition task required retrieval of spatial or object information. Our data suggest that the anterior hippocampus is involved in forming memory representations of novel environments whereas the PPA plays a broader role in scene processing that is not limited to encoding.

(End of symposium)

Session B - Saturday 22 July 11.20-1 pm

Where is the word length effect? Right here! A reply to Bachaud-Levi et al. (1998).

Antje S Meyer1, Ardi Roelofs2, and Pim Levelt2

1. University of Birmingham

2. Max Planck Institute for Psycholinguistics, The Netherlands

The model of word production proposed by Levelt, Roelofs, and Meyer (e.g., BBS,1999) predicts that it should take speakers longer to initiate long words than shorter ones. This is because the model assumes that successive syllables of a word are generated in sequence and that speakers usually generate all syllables of a word before speech onset. In several earlier studies this prediction was borne out, but Bachoud-Levi et al. (JML, 1998) failed to obtain any evidence for a word length effect on speech onset latencies. They concluded that the model had to be modified.

In our experiments, Dutch and English speakers named objects with monosyllabic or disyllabic names, which were equally easy to recognize. When mono- and disyllabic targets were presented in mixed blocks, there was no length effect. However, when the materials were blocked for length, or when the participants were strongly encouraged to react fast, length effects were obtained. We argue that these results are consistent with our model's assumptions about the time course of phonological encoding and that processing time difference can easily be concealed (in mixed blocks and when no speed instruction is given) because of the participants' setting of a response deadline.

A critical test of the changing routes versus changing deadlines debate in print to sound translation

Ilhan Raman1, Bahman Baluch1 and Derek Besner2

1. Middlesex University

2. University of Waterloo

The word frequency effect in speeded naming of single words can be eliminated by mixing nonwords together with the words (e.g., Baluch & Besner, 1991; Hudson & Bergman, 1985) Two different accounts have been proposed. In one account, the reader shifts from relying on the addressed routine (which produces a word frequency effect) when only words are present to relying on the assembled routine (sublexical spelling-sound correspondences which do not produce a word frequency effect) when nonwords are mixed in with the words (Baluch & Besner, 1991). A different account proposes that the reader does not shift routes, but instead trys to homogenize the point in time at which an articulation is released, thus pulling the release point for high frequency words towards the release point for nonwords, consequently eliminating the word frequency effect (Lupker, Brown & Columbo, 1997).

These two different accounts are pitted against each other in a series of single-word naming experiments utilising the transparent Turkish orthography in which a robust frequency effect was previously reported (Raman, Baluch & Sneddon, 1996). The critical manipulation involves mixing words together with nonwords which are as fast as words to name, since the "changing routes" account predicts that the word frequency effect will be eliminated whereas the "changing deadlines" account predicts that a word frequency effect will be preserved. The results are inconsistent with the "changing routes" account but consistent with the "changing deadlines" account insofar as naming in a transparent script is concerned. The implications of these findings for understanding reading at the single word level in various writing systems are discussed.

Baluch, B. & Besner, D. (1991). Visual word recognition: Evidence for strategic control of lexical and nonlexical routines in oral reading. Journal of Experimental Psychology: Learning, Memory and Cognition, 17 (4), 644-652.

Hudson, P.T.W. & Bergman, M.W. (1985) Lexical knowledge in word recognition: word length in naming and lexical decision tasks. Journal of Memory and Language, 24, 46-58.

Lupker, S. J., Brown, P. & Colombo, L. (1997). Strategic control in a naming task: Changing routes or changing deadlines? Journal of Experimental Psychology: Learning, Memory and Cognition, 23 (3), 570-590.

Raman, I., Baluch, B. & Sneddon, P. (1996). What is the cognitive system's preferred route for deriving phonology from print? European Psychologist, 1 (3), 221-227.

Instantiating counter-intuitive frequency effects within a distributed memory model of naming

Thomas M Spalek. and Steve Joordens,

University of Toronto

Masson has proposed a distributed memory model as an account of the memory retrieval process underlying word recognition (1991,1995). Piercey & Joordens (1999) augmented this model with a random walk decision process in an attempt to map this general word process onto lexical decision performance. The current work examines whether a modified version of this model can also explain performance in a naming context. We will first begin by briefly summarizing some of the naming results and our modified deadline model that we presented at last year's CSBBCS meeting. Specifically, we emphasize the counter-intuitive frequency effects found when you mix monosyllabic and trisyllabic items. We then describe the results of simulations designed to instantiate our deadline assumptions within Masson's distributed framework.

Repetition blindness for words but not non-words

Jamie I D Campbell and Vanessa Hernberg

University of Saskatchewan

Repetition blindness (RB) is the failure to recognize repetition of an item in a rapid visual series of stimuli. We examined RB for English words and pronounceable non-words. Sixty undergraduates each received 192 trials. For each trial, 6-13 items (mean = 9) appeared for 86 ms sequentially in the same location. Half were randomized sequences and half formed meaningful sentences. Each sequence contained two word targets or two non-word targets separated by one to three items. Targets were brighter than the non-target words. Following the sequence, three response items appeared. For repetition trials these included the repeated target and two not-presented fillers; for non-repetition trials these included the two targets and one filler (Arnell & Jolicoeur, 1997). Participants indicated whether there were had been 0, 1 or 2 occurrences of each response item. Analysis of a-prime, percentage correct, and hits-false alarms all indicated RB for words, but no RB for non-words.

Putting automaticity in context: Reducing the Stroop effect

Colin M. MacLeod

University of Toronto at Scarborough

In a recent series of papers, Besner and Stolz have argued that "simply coloring a single letter instead of the whole word eliminated the Stroop effect." They have used this outcome to argue against the concept of universal automaticity of word reading, claiming that reading cannot be automatic if it can be prevented. I will report experiments demonstrating that the Stroop effect is not generally eliminated by this manipulation, although it is reliably decreased across both manual and vocal response modes, and relative to different types of control items. The word definitely is read in the single-letter-colored condition, but its impact on color naming is reduced. Although not diagnostic of a change in automatic reading, this reduction is an important clue to understanding the source(s) of Stroop interference.

Session C - Saturday 22 July 11.20-1 pm

An oblique effect in aesthetics: Homage to Mondrian (1872-1944)

Richard Latto, Douglas Brain and Brian Kelly

University of Liverpool

The effect of the orientation of Mondrian's paintings on their aesthetic appeal was examined. Eight paintings, four with horizontal/vertical frames in the original and four with oblique frames, were presented in eight different orientations and rated for aesthetic appeal on a 7-point scale.

There was a stronger preference for pictures presented so that their component lines were horizontal and vertical than for pictures presented with their component lines in an oblique orientation. In addition, subjects showed a preference for the original orientation, perhaps because rotation changes the lateral balance of the paintings as well as the orientation of the component lines. There was no overall preference for one frame orientation over another, but there was an interaction between frame orientation and component orientation, resulting in a preference for paintings where the components were parallel to the surrounding frame.

It is suggested that the aesthetic oblique effect reported here is related to the oblique effect in orientation perception and the privileged access which horizontal and vertical lines have to the visual system. This offers support for one possible mechanism underlying aesthetic judgements of abstract patterns: we find pleasing those stimuli which are closely tuned to the properties of the human visual system (Latto, 1995).

Latto, R. (1995). The brain of the beholder. In R. L. Gregory, J. Harris, P. Heard & D. Rose (Eds.), The Artful Eye (pp. 66-94)). Oxford, U.K.: Oxford University Press.

Five plus two equals yellow: Concept-driven photisms in digit - colour synaesthesia

Mike J Dixon, Daniel Smilek, Cera Cudahy and Philip M Merikle

University of Waterloo

When the synaesthete C views black digits, each number elicits a photism - a visual experience of a specific colour. Cytowic (1993; 1997) has claimed that photisms differ from imagery in their automaticity and their reliance on an external stimulus to elicit them. To assess the automaticity of C's photisms, digits were displayed in colours incongruent or congruent with her photism colours. Despite attempts to ignore digits, C's colour naming reaction times were significantly slower for incongruent trials than congruent trials. Finally, we triggered automatic photisms without ever physically presenting an inducing stimulus. C was shown arithmetic problems (e.g. 5 + 2) followed by a colour patch that she had to name. Naming times were significantly slower when colours were incongruent with C's photisms for the answers to problems than when colour patches were congruent. We conclude that C's photisms are automatic, but can be induced by activating the concept of digits.

Repetition priming for moving faces.

K Lander and V Bruce

University of Stirling

Recent experiments suggest that seeing a familiar face move adds additional ‘dynamic’ information for the viewer, useful in the recognition of identity (see Knight & Johnston, 1997; Lander, Bruce & Hill, 1999 in press; Lander, Christie & Bruce, 1999). This finding is consistent with the idea that familiar movement patterns – either of faces generally, or specific faces, help in their recognition. It is unclear, however, how such information may be stored in memory. One possibility is that the movement patterns are quite distinct from the static form-based representation. Alternatively, the visual representations of familiar faces may intrinsically incorporate dynamic information (cf. dynamic representations, Freyd, 1993).

Here we describe a number of experiments which aim to explore these theoretical possibilities, using a repetition priming technique. Repetition priming is the advantage demonstrated at test when a to-be-recognised item (in this case a face) is encountered earlier, at some time prior to the test event. Results suggest that a moving image primes more effectively than a static image (Experiment 1), even when the same static image is shown in the prime and test phases of the experiment (Experiment 2). This robust priming effect is not simply due to the structured sequence of images contained in a moving sequence (Experiment 3). Results are discussed within the framework of current models of face recognition.

Freyd, J.J. (1993). Five hunches about perceptual processes and dynamic representations. In Meyer, D. & Kornblum, S. (Eds), Attention and Performance, XIV: Synergies in Experimental Psychology, Artificial Intelligence and Cognitive Neuroscience - A Silver Jubilee, 99-120. Cambridge, MA: MIT Press,

Knight, B. & Johnston, A. (1997). The role of movement in face recognition. Visual Cognition, 4(3), 265-273.

Lander, K., Christie, F. & Bruce, V. (1999). The role of movement in the recognition of famous faces. Memory & Cognition, 27(6), 974-985.

Lander, K., Bruce, V. & Hill, H. (1999, in press). Evaluating the effectiveness of pixelation and blurring on masking the identity of familiar faces. Applied Cognitive Psychology.

Getting to know we learn new faces

Vicki Bruce and Zoe Henderson

University of Stirling

Recent research has highlighted the differences between the visual representations of familiar and unfamiliar faces. Unfamiliar faces are stored in an image-specific way, allowing only limited generalisation to different viewpoints, expressions or lighting conditions. Familiar face recognition, in contrast, is robust over many such variations in viewing conditions. Representations for familiar faces are weighted more towards internal features compared with unfamiliar faces. Angeli, Bruce and H. Ellis (ECVP, 1999) described preliminary results demonstrating that the shift in processing from external to internal face features can be used as an index of familiarisation as participants learn a series of individuals on video over several days, and at a recent EPS meeting, O'Donnell and Bruce described experiments suggesting that it was particularly the eyes which benefit from familiarisation. In this paper we describe the results of a new experiment where we examine the process of familiarisation when participants learn new facial identities from moving compared with multiple static images.

Individual differences in the development of face recognition: Support for a maturational change at age eight.

P A McMullen, P Dunham, and F Dunham

Dalhousie University, Halifax

Carey & Diamond (1977) suggested that before the age of 8, children recognize a face by its individual features. After this age, a whole-face (holistic) strategy has developed. Boys and girls aged 7 and 11 performed tests of Mooney faces (a test of holistic processing), Embedded figures (a test of feature-based processing) and upright and inverted face recognition. Children aged 7, successfully recognized upright faces but were at chance at the recognition of inverted faces. More importantly, upright face recognition correlated with Embedded figures performance (r= 0.56, p<0.01) and not with Mooney face performance. Children aged 11, recognized upright and inverted faces above change levels. Notably, their results showed the opposite pattern to those of the younger children. Upright face recognition now correlated with Mooney face performance (r= 0.45, p<0.05) and not with Embedded figures performance. The maturational hypothesis of face recognition was tested from an individual differences approach and supported.

Session A - Saturday 22 July 2-4 pm

Correlation, consistency, and age of acquisition effects in adult lexical processing

Andy Ellis1, Antonina Scarna1, Josephine Monaghan1 and Matt Lambon Ralph2

1. University of York

2. MRC Cognition and Brain Sciences Unit, Cambridge

Ellis and Lambon Ralph (in press) have proposed an explanation of age of acquisition effects in adult word recognition and production in terms of the behaviour of distributed memory networks trained by backpropagation. Such networks are widely used to simulate aspects of adult lexical processing. Ellis and Lambon Ralph showed that if a network is trained from the outset on one set of "early" patterns, then later patterns are added, with all of the patterns now being trained in a cumulative and interleaved fashion, then performance of the network remains superior on early compare with late patterns, even after extensive further training. This effect cannot be explained in terms of simple differences in the cumulative frequency of presentation of the patterns. Age of acquisition effects, like frequency effects, are a natural and probably inescapable property of these networks. We present new simulations which show that the age of acquisition is modulated by the similarity or correlation between early and late patterns. When they have low correlations, network structure built up during training on the early patterns is unhelpful for the assimilation of later patterns which fail to attain representations comparable to those of the early patterns. In contrast, when the correlation between early and late patterns is high, the late patterns can take advantage of the network structure established by the early patterns and age of acquisition effects are reduced. We present old and new data showing that age of acquisition effects are larger when the correlation between early and late items is low (naming objects in English and Italian; reading Japanese kanji; reading aloud exception words in English) and smaller or nonexistent when the correlation is high (reading aloud consistent words in English, and reading aloud in Italian).


Ellis, A.W., & Lambon Ralph, M.A. (in press). Age of acquisition effects in adult lexical processing reflect loss of plasticity in maturing systems: Insights from connectionist networks. Journal of Experimental Psychology: Learning, Memory & Cognition.

Homophone effects in lexical decision: An examination of a feedback account

Penny M Pexman1 and Stephen J Lupker2

1. University of Calgary

2. University of Western Ontario

Pexman, Lupker, and Jared (in review) observed the standard finding of longer response times in lexical decision tasks (LDT) to homophones (e.g., MAID/MADE) than to non-homophonic control words. Pexman et al. attributed this homophone effect to the nature of the feedback from phonology to orthography for homophones (i.e., one phonological code feeds back to two orthographic codes). In these experiments we tested three predictions of this feedback account: (a) although homophone effects are observed in LDT, regularity and homograph effects should not be, since most exception words (e.g., HAVE) and homographs (e.g., WIND) do not create the same feedback problems, (b) using pseudohomophones as the nonwords should affect only the size of the homophone effect and (c) in a phonological LDT ("does it sound like a real word?") regularity and homograph effects should be observed but homophone effects should not. The results supported our predictions for all but a subset of our (poorer) readers.

Lexical and relation-based influences on the interpretation of noun-noun phrases

Christina L Gagne

University of Western Ontario

Recent exposure to a similar combination influences the interpretation of a noun-noun phrase by increasing the availability of the phrase¹s lexical entries and (in some cases) the availability of the relation used to interpret the combination. The data show that the amount of lexical and relation priming obtained depends on whether the modifier or head noun is in common between the prime and target. The head noun prime yields more lexical priming than does the modifier prime. In contrast, relation priming is only obtained from the modifier prime. For example, oil treatment is easier to interpret when it has been preceded by a combination (e.g., oil moisturizer) using the same relation than when preceded by a combination (e.g., oil accident) using a different relation. The data are used to evaluate current theories of conceptual combination.

Phonological mediation in a semantic discrimination task: Evidence from French

Isabelle Gonthier, Alain Desrochers and Dominique Landry

University of Ottawa

Lexical phonology and semantic was investigated in word recognition. Word triplets were presented to readers who were asked to decide which of the first two words, a target word (e.g., SABLE) or a distractor, was semantically related to a test word (e.g., PLAGE). The distractor was either homophonic with a semantically related word (e.g., MÈRE which can evoke MER) or a visually similar control stimulus (e.g., MOIS). The imagery value of word triplets was manipulated. The results indicated that a) response latencies were slower when the distractor evoked a semantically related homophone than a visually similar control; b) responses to high-imagery stimuli were faster than those to low-imagery stimuli; c) no interaction was detected between the homophonic and imagery effects. These results suggest that evoked imagery does affect word meaning processing but does not constrain phonological mediation in this task.

Separating form and semantics from morphology: Evidence from cross modal masked priming.

William Marslen-Wilson and Mike Ford

MRC Cognition and Brain Sciences Unit, Cambridge

A recurring issue in the study of the English mental lexicon is to separate out evidence for morphological structure in underlying representation from the effects of form overlap (orthographic or phonological) and semantic overlap between morphologically related items (e.g., darkness/dark, rebuild/build). A partial solution to this is provided by masked priming, which can show robust morphological effects in the absence of semantic effects. However, conventional masked priming is strongly affected by orthographic overlap between primes and targets. Here we report a new masked priming paradigm, where a masked visual prime is followed not by a visual target but by an auditory target. Here there is no direct form overlap between prime and target. In two experiments we find significant identity priming (where prime and target are the same word), and significant priming between morphologically related prime/target pairs, such as adaptable/adapt. This has implications both for claims about the role of morphology in the mental lexicon, and for our understanding of the underlying mechanisms of masked priming.

The rise and fall of frequency and imageability: Noun and verb production in semantic dementia

Helen Bird1, Matthew A Lambon Ralph1, Karalyn Patterson1 and John R Hodges1,2

1. MRC Cognition and Brain Sciences Unit, Cambridge

2. University Neurology Unit, Addenbrooke's Hospital, Cambridge

This study examines the impact of progressive degeneration of conceptual knowledge on the content words used in connected speech elicited using the Cookie Theft picture description (Goodglass & Kaplan 1983). We began with an analysis of control subjects' descriptions with regard to word types and their frequency and imageability. Because the impairment of conceptual knowledge in semantic dementia is graded by concept familiarity, we created a model of a standardised normal Cookie Theft description that was then progressively degraded by the systematic removal of lower bands of word frequency. We drew two main predictions from this model: reduced availability of the lower bands of word frequency should result in (a) an apparent deficit for noun retrieval in relation to verb retrieval, and (b) an apparent reverse imageability effect. Results from a longitudinal study, in which three patients with semantic dementia each described the Cookie Theft picture on three occasions during the progression of their disease, confirmed these predictions. An additional cross-sectional analysis, adding narratives from a larger number of cases, demonstrated that the decline in ability to produce suitable words for the picture description is closely related to the extent of semantic impairment as measured in tests of word comprehension and production. Both verbs and nouns are affected by the degradation of semantic memory; the fact that the impairment to noun production is manifested earlier and more catastrophically may be attributed to the relatively lower frequency of these terms.

Goodglass, H. & Kaplan, E. (1983). The Assessment of Aphasia and Related Disorders. (2nd edition). Lea & Febiger, Philadelphia.

Session B – Saturday 22 July 2-4 pm

The name game: Using retrieval practice to learn the names of group members.

Peter E Morris1 and Catherine O Fritz2

1. Lancaster University

2. Bolton Institute

In medium sized groups such as classes it is often desirable that the members become acquainted with one another. To this end various methods of introducing group members are used with questionable degrees of success. One method of introduction, the Name Game, is based upon the principles of retrieval practice and anecdotal evidence supports its effectiveness. We compared Simple and Elaborate versions of the Name Game with a widely used method - Pairwise Introductions - and found that the Name Game participants were much better at remembering the full names of the members of their groups after 30 minutes, 2 weeks and 11 months. The effectiveness of the two Name Game methods did not differ. In a second experiment, we tested the contribution of retrieval practice by comparing the two versions of the Name Game with a procedure that matched them in number of repetitions and time spent on the task. Again, the Name Games were much superior, illustrating the benefits of retrieval practice in general and the Name Game technique in particular.

Interrupting recognition memory: Tests of the increment-to-familiarity account of the revelation effect

Marty W Niewiadomski and William E Hockley (Presenter: William E. Hockley)

Wilfrid Laurier University

The revelation effect is a puzzling phenomenon in which items on a recognition test are more likely to be judged old when they are immediately preceded by a problem solving task such as anagram solution. We evaluated Westerman and Greene's (1998) increment-to-familiarity account of this phenomenon in a series of experiments designed to examine the temporal and cumulative effects of the preceding task on the recognition probe. We found comparable revelation effects when the probe was preceded by (a) a single anagram versus two anagrams, (b) either combination of a single anagram and an arithmetic task sequence, and (c) one versus two arithmetic tasks. The results do not support the increment-to-familiarity account of the revelation effect, but are consistent with Hicks and Marsh's (1998) decision-based interpretation, in which it is assumed that the preceding task adds noise to working memory leading subjects to adopt a more liberal recognition decision criterion.

Feature frequency in recognition from short-term memory: A challenge for current theory

D J K Mewhort and E E Johns

Queen's University

Current global models of recognition memory assume a fundamental role for frequency at the feature level: The more often the constituent features of a test item have occurred during study, the easier it is to call OLD and the harder it is to call NEW. We manipulated the featural composition of artifical stimuli in a short-term recognition memory task. Performance on distractor items deteriorated with the number of constituent features studied, but, contrary to prediction, did not vary with the number of times a presented feature had occurred in the study set. Conversely, performance on targets improved when constituent features had occurred in other items within the study set. We conclude that different kinds of evidence underlie OLD and NEW responses and that recognition decisions are not correctly modelled with a signal-detection mechanism. Note that the mirror effect is not a source of difficulty for our position.

Estimating monitoring, bias and retrieval

Philip A Higham

University of Northern BC

Research is described in which a Type 2 signal detection model (SDM) is applied to data from various memory tests, yielding estimates of report bias (the tendency to offer answers to questions on a memory test or the tendency to offer information during a memory interview), monitoring (a meta-memory parameter indexed by the tendency to offer correct information and withhold incorrect information) and retrieval (the proportion of correct information on a memory test once all information is obtained). The research indicates that the estimates are affected, and sometimes dissociated, by variables (e.g., a target word’s strength-of-association to a retrieval cue) in logical and sensible ways that are consistent with current models of memory. The research also indicates that the SDM model can be applied to real-world memory problems as well, such as how best to enhance performance on aptitude tests (e.g., SAT) or on multiple-choice exams.

The role of source memory in setting the criteria for signal decision processes

C A G Hayman1, J Fugelsang2, J Cofell3 and R P Cribbie4

1. Lakehead University

2. University of Saskatchewan

3. University of Western Ontario

4. University of Manitoba

In a series of experiments testing both words and pictures as to-be-remembered material, we found that subjects recall more about the source of a memory event when they gave a "REMEMBER" rather than "KNOW" response in a prior test of recognition.  Such a finding is inconsistent with the assumption of an unidimensional vector of memory strength required in signal detection based reinterpretations (Donaldson , 1996; Hirshman & Master 1997) of dual-process views of recognition memory (Gardiner & Java, 1993; Jacoby & Dallas, 1981; Mandler, 1980).  Because increase in source memory require increases in the multidimensionality of memory, we suggest that the unidimensional vector of memory strength hypothesized in these signal detection models is at best an approximation of a multidimensional memory signal, and that criteria used in signal detection models of "REMEMBER" and "KNOW" response may be set relative to the availability of memory information for source.

Implicit/explicit memory versus analytic/nonanalytic processing: Re-thinking the mere exposure effect

Bruce W A Whittlesea and John R Price

Simon Fraser University

In studies of the mere exposure effect, rapid presentation of training items can increase liking without producing accurate recognition. The effect on liking has been explained as a misattribution of fluency caused by prior presentation. However, fluency is also a source of feelings of familiarity. It is therefore surprising that prior experience can produce liking without also causing familiarity-based recognition. We suggest that when study opportunities are minimal and test items are perceptually similar, people adopt an analytic approach, attempting to recognize distinctive features. That strategy fails because rapid presentation prevents effective encoding of such features; but it also prevents people from experiencing fluency and a consequent feeling of familiarity. We suggest that the liking-without-recognition effect results from using an effective (nonananlytic) strategy in the pleasantness task, but an ineffective (analytic) strategy in the recognition task. We suggest that explanations of the mere exposure effect based on a distinction between implicit and explicit memory are unnecessary.

Session C – Saturday 22 July 2-4 pm

Unlearned and learned behaviour of bumble bees in the absence of reward

V Simonds and C M S Plowright

University of Ottawa

This work features the development of a unique methodology to test unlearned and learned behaviour of bumble bees in the absence of reward. Two questions were asked: What attracts newly emerged bees to flowers? Does the attraction habituate over time? Two experiments were designed to test these questions. 1A: Bees made choices in a 12-arm maze between four colours and preferred yellow or blue over white or red. 1B: Choices for yellow decreased from the first to the second and third testing sessions. 2A: Bees preferred a radial pattern over a concentric pattern or a plain disc. 2B: Choice frequencies for radial decreased over the first two sessions, but increased for a novel stimulus in the third session. Results support previous findings in unlearned responses to pattern and colour, and support the previously neglected notion that bumble bees learn in the absence of reward.

The discrimination of structure

David N George and John M Pearce

Cardiff University

The solution of certain, configural, discriminations depends upon learning about the significance of combinations of stimuli, rather than of individual stimuli. According to elemental theories of conditioning (e.g., Wagner & Rescorla, 1972) mastery of a configural discrimination depends upon associations involving unique cues that are created by thecombination of two or more stimuli. Configural theories of conditioning (Pearce, 1994), on the other hand, explain the solution of these discriminations by allowing the features of a pattern of stimulation collectively to signal the outcome of a trial. The experiment to bedescribed demonstrates that pigeons can solve a complex configural discrimination which both the Wagner-Rescorla (1972) theory, and the theory of Pearce (1994), predict will be insoluble. The discrimination depends upon an appreciation of the structure of the signals for reward and nonreward. Various ways in which elemental and configural theories of conditioning might be modified to explain our results will be considered.

Pearce, J. M. (1994). Similarity and discrimination: A selective review and a connectionist model. Psychological Review, 101, 587-607.

Wagner, A. R., & Rescorla, R. A. (1972). Inhibition in Pavlovian conditioning: Application of a theory. In M. S. Halliday & R. A. Boakes (Eds.), Inhibition and Learning. San Diego, CA: Academic Press.

Losing a conditioned aversion to a taste: Extinction or counter-conditioning?

Bob Boakes and Paul Whitfield

University of Sydney, Australia

Recent findings from human evaluative conditioning and from flavour preference conditioning in rats prompted re-examination of Garcia's original claim that conditioned taste aversions are unusually resistant to extinction. Evidence against this claim has been based on giving repeated single-bottle exposure to an averted taste to thirsty rats, a procedure that typically leads to loss of the aversion. However, it also involves pairing the taste with relief of fluid deprivation and such pairing could produce counter-conditioning of the aversion. This possibility was examined by orally perfusing rats with the averted taste either just before (Thirsty condition) or just after (Sated condition) they received their daily period of access to water. Repeated exposure to the taste using this method was found to produce more rapid loss of the taste aversion in the Thirsty than in the Sated condition. The results suggest that counter-conditioning does play a role in the loss of a conditioned aversion and that, in the absence of this factor, Garcia's claim might well be correct.

The anti-emetic drug, ondansetron, interferes with lithium-induced conditioned rejection reactions, but not lithium-induced taste avoidance.

Cheryl L Limebeer and Linda A Parker

Wilfrid Laurier University

Conditioned rejection reactions displayed in the taste reactivity test are exclusively by treatments that elicit nausea.  The present experiments demonstrated that pretreatment with the anti-nausea agent, ondansetron, interfered with both the establishement and the expression of conditioned rejection reactions.  Ondansetron did not interfere with lithium-induced taste avoidance in either a one-bottle or a two-bottle test.  In fact, when rejection reactions were measured during a consumption test, ondansetron selectively attenuated rejection reactions without modifying consumption.  These results suggest that conditioned rejection reactions, but not conditioned taste avoidance, reflect nausea in rats that can be attenuated by ondansetron pretreatment.

Do squid make a visual language on their skin? The case of the Zebra display.

Jennifer A Mather

Department of Psychology & Neuroscience, University of Lethbridge

In the 1980s, Martin Moynihan suggested that the production of a combination of local components of skin pattern changes on the skin of the squid indicated an ability to form a visual language. But production of a range of components, no matter how wide or how flexible in combination, does not prove a language. The present study focuses on the variation in one display, the dark-and-pale striped Zebra. Observations of over 1200 instances of Zebra displays of Sepioteuthis sepioidea in the waters off the coast of Bonaire confirmed a huge variation in this display. Zebra could be produced on part or all of the body surface, by males or females, in different intensities and to a variety of targets. Despite the variety, the display appeared to be essentially an agonistic signal of internal state. A quantitative score of signal intensity could be calculated and was often correlated with success in intraspecific interactions. Nevertheless, the intensity was not always correlated with dominance, the signal was often directional and it was also modified to a ritualized Formal Zebra contests between adult males. This display, while of a sophistication and range of use that fulfils Moynihan's prediction of communicative sophistication, does not yet appear to suggest a visual language.

The relation between stress and the social organization of wolves and other wild canids

Peter J McLeod, Simon Gadbois and Will Moger

Acadia University, Nova Scotia

We have been relating detailed data on social behaviour of a captive wolf pack to their levels of the stress responsive hormone cortisol, assayed non-invasively from urine-contaminated snow. These data show significant differences among individuals that could be due to social rank and/or challenges to their rank. Adrenal pathology due to a tumour is a possible explanation in one case as well. Inconsistent correlations between cortisol and rank have been reported across a variety of wild canids and other social mammals. We can account for some of these differences by relating them to differences in social organization and reproductive strategies. Theoretical considerations of the degree to which species are monogamous and cooperative breeders are particularly relevant in accounting for existing data.