Elsevier

Neural Networks

Volume 18, Issue 4, May 2005, Pages 353-369
Neural Networks

2005 Special Issue
The interaction of attention and emotion

https://doi.org/10.1016/j.neunet.2005.03.005Get rights and content

Abstract

We analyse emotions from the viewpoint of how emotion and attention interact in the brain. Much has been learnt about the brain structures involved in attention, especially in vision. In particular the manner in which attention functions as a high-level control system, able to make cognitive processing so effective, has been studied both at a global level by brain imaging (fMRI, PET, MEG and EEG), at a local single cell level in monkeys and lower animals, and computationally by a variety of models. The manner in which emotions impinge on this attention control system is not so well analysed, although numerous new results are now emerging from using the same tools. Here we use an engineering control approach to attention to model it in a global manner but with relatively sure local foundations at singe neuron level.

The manner in which emotional value (as coded in amygdale and orbito-frontal cortex) can interact with the attention control circuitry is analysed using results of various experimental paradigms. A general model of this interaction is first developed and tested against a list of paradigms, and then more detailed computations are performed using more specific features of the attention control system and the limbic value coding. These computations are completed by a simulation of the emotional attentional blink, a demanding paradigm for any model of attention alone, but made more so by the presence of emotional value codes for stimuli. We conclude the paper with a general discussion of further avenues of research.

Introduction

The commonplace view of emotions is that they provide the ‘colouring’ or ‘flavouring’ of activity in everyday life. Without emotions, life is experienced as having little meaning or purpose, and the pleasures that are derived from rewarding experiences are considerably reduced. Emotionless actions are often related to machines, which are executing a sequence of pre-programmed commands. While this commonplace perception of the role emotions play in everyday life is pertinent it is somewhat limited in scope. For emotions not only ‘colour’ human behaviour but to a large extend guide it as well.

In order to understand how such guidance can arise in the brain, it is necessary to consider what is being guided. In general it is expected to be a response, especially between competing possibilities. The overarching control system for such a response, and for the selection of stimuli to which any response is to be made, is the faculty of attention. This faculty is used especially in the early stages of response, before any over-learning occurs (leading to automatic responses), so plays a most crucial role in ensuring efficiency and survivability. Thus we consider that in order to understand emotions it is necessary to build models of how emotions guide attention itself. It is that logic that has caused us to investigate the interaction between attention and emotion, as the title of our paper states.

Attention is a faculty of many animal species, used to reduce the flow of information into the sensory system of the brain by enhancing the relevant or important components of the input stream while eliminating the distracting ones. This operation is guided by the goals, sited mainly in the prefrontal cortex, that maintain the template for these relevant or important components. What determines the relevance or importance of incoming stimuli according to which attention is applied and actions in response to the attended stimuli are generated?

The answer to this important question is that it is the emotional content of stimuli presented to the sensory system that are the principal indicator of the importance of these stimuli. Thus emotional content can modify and update the goals and consequently alter the direction of attention to the presented stimuli. Emotions and goals are strongly intertwined in the sense that the immediate relevance of any stimulus to a goal defines the emotionality of the stimulus. The relationship between emotions and the personal goals and concerns of individuals is often suggested to be the basis for emotion elicitation and differentiation by appraisal theorists (see Scherer, 1999 for a review). For instance the emotional tag of fear can be attached to a threatening stimulus in so far as the latter can potentially impede the goal of survival. Another example is the emotional tag of happiness that can be assigned to any stimulus that advances the goal of well-being. In a similar fashion numerous emotional tags can be given to stimuli that promote or hinder the attainment of goals ranging from basic individual survival goals to more complex social interaction goals. This vast range of emotions and the related goals is not likely to have been formed concurrently. Rather, emotions evolved from very simple mechanisms that ensured harm avoidance and attainment of vital physical resources into more complex mechanisms that guide complex social behaviour. This evolution of emotions may in fact be reflected in the brain systems that generate them, with emotions linked to survival arising from evolutionarily old brain systems, while emotions linked to complex social behaviour developing in phylogenetically more recent brain areas. Thus the more primitive emotions would be expected to be elicited by more primitive aspect so of the environment, and only at higher levels of evolution would complex classification of stimuli have had related emotions associated with them.

So far we have avoided a formal definition of emotions and restricted ourselves to illustrating only some basic features of them. However, a distinction should be made between emotions as labels attached to stimuli presented to the sensory system and to the experience of emotions or the so-called ‘feelings’ (LeDoux, 2000). Feelings can be regarded as states of emotional consciousness, bearing all the relevant characteristics of conscious states. Thus feelings echo the phenomenal experience of emotions or the ‘what it is like’ as in ‘what it is like to be sad’. The neural basis of the conscious experience of emotions has been regarded by some scientists as unimportant compared to the much more tractable neural signatures of emotions themselves (LeDoux, 2000), but others consider emotional experience to be a crucial issue in understanding the role emotions play in human behaviour (Lambie and Marcel, 2002, Fragopanagos and Taylor, 2004, Fragopanagos and Taylor, 2004). However, such a pursuit lies outside the scope of this paper insofar as we here investigate the manner in which emotions may (or may not) attract attention to sensory stimuli, thus selectively enhancing their perception. Given that such emotion-laden stimuli are attended and fully processed, more complex processes relating these stimuli with long-term emotional memories can then lead to emotional experience (although we will not deal with these processes here).

In what follows in this paper we will first discuss the nature of emotions and, particularly the concept of their automatic processing, and then look at how emotions may interact with the attentional selection mechanisms based on the experimental evidence. We will then present a neural model of the interface of emotions and attention and apply this model to explain qualitatively and quantitatively a series of experimental paradigms. The paper finishes with conclusions and discussion of further directions of associated research.

Section snippets

The automatic view of emotions

Before attempting to put together any model of the interaction of attention and emotion, we must address an important question about the nature of emotions, that is, whether or not emotions are encoded ‘automatically’ and what this ‘automatic’ encoding could mean. Indeed the concept of ‘automaticity’ when applied to brain processing is not as straightforward as when applied for instance to engineering, where it is most often encountered anyway. In engineering ‘automatic’ implies an operation

Guidance of attention by emotion

So far we have reviewed the evidence for the existence of an automatic mechanism for the registration of the emotional value of stimuli presented to the sensory systems of the brain, under the assumption that such a mechanism would be fast, possibly not requiring conscious awareness and presumably posing minimal demand for attentional resources. Given that such a mechanism exists we now turn to examine the evidence and possible mechanisms by which these automatically extracted emotional cues

General model of the interaction of attention and emotion

Brain imaging and brain deficit results in depressives indicate the division of processing into: a ventral network for emotion, and a dorsal one for cognition. The imbalance between the two networks, driven by excessive emotional activity, leads to reduction of the cognitive activity and an excess of limbic activity. This feature leads to the question as to the nature of the interaction between emotion and attention: are they competing ‘attention-type’ systems, or is attention the main control

Simulation of Yamasaki et al. (2002).

The paradigm of Yamasaki et al. (2002) had various types of stimuli presented to subjects lying in the bore of the magnet in an fMRI machine. Standards consisted of squares of varying sizes and colours, targets consisted of circles of varying sizes and colours, and emotional distracters consisted of aversive pictures that included unpleasant themes of human violence, mutilation, and disease. Finally, neutral distracters consisted of pictures of ordinary activities. The task was to press a

Conclusions

In this paper we have investigated the interaction of emotion and attention through a careful review of the psychological and neuroimaging literature as well as through explicit modelling of various attention/emotion paradigms by means of a model developed by extending the CODAM model for attention to include limbic and paralimbic structures. More specifically, we presented evidence from the literature supporting the notion that the emotional value of stimuli or events presented to the

Acknowledgements

We would like to acknowledge the EC under the project ERMIS for support while this work was being done, as well as our colleagues in ERMIS, especially R Cowie, E Douglas-Cowie and the QUB team and S Kollias and the NTUA team.

References (57)

  • R.D. Lane et al.

    Neuroanatomical correlates of pleasant and unpleasant emotion

    Neuropsychologia

    (1997)
  • J.S. Morris et al.

    Parallel neural responses in amygdala subregions and sensory cortex during implicit fear conditioning

    Neuroimage

    (2001)
  • S. Ortigue et al.

    Electrical neuroimaging reveals early generator modulation to emotional words

    Neuroimage

    (2004)
  • L. Pessoa et al.

    Attentional control of the processing of neural and emotional stimuli

    Brain Research. Cognitive Brain Research

    (2002)
  • G.J. Quirk et al.

    Fear conditioning enhances different temporal components of tone-evoked spike trains in auditory cortex and lateral amygdala

    Neuron

    (1997)
  • E.T. Rolls

    The functions of the orbito-frontal cortex

    Brain and Cognition

    (2004)
  • K.M. Stormark et al.

    Conditioned emotional cueing of spatial attentional shifts in a go/no-go RT task

    International Journal of Psychophysiology

    (1997)
  • M.H. Tabert et al.

    Differential amygdala activation during emotional decision and recognition memory tasks using unpleasant words: An fMRI study

    Neuropsychologia

    (2001)
  • A.M. Treisman et al.

    A feature-integration theory of attention

    Cognitive Psychology

    (1980)
  • P. Vuilleumier et al.

    Neural response to emotional faces with and without awareness: Event-related fMRI in a parietal patient with visual extinction and spatial neglect

    Neuropsychologia

    (2002)
  • P. Vuilleumier et al.

    Effects of attention and emotion on face processing in the human brain: An event-related fMRI study

    Neuron

    (2001)
  • A.K. Anderson et al.

    Lesions of the human amygdala impair enhanced perception of emotionally salient events

    Nature

    (2001)
  • M. Beauregard et al.

    Neural correlates of conscious self-regulation of emotion

    Journal of Neuroscience

    (2001)
  • S. Campanella et al.

    Categorical perception of happiness and fear facial expressions: An ERP study

    Journal of Cognitive Neuroscience

    (2002)
  • L. Carretie et al.

    Automatic attention to emotional stimuli: Neural correlates

    Human Brain Mapping

    (2004)
  • M. Corbetta et al.

    Control of goal-directed and stimulus-driven attention in the brain

    Nature Reviews. Neuroscience

    (2002)
  • B. de Gelder et al.

    Unseen stimuli modulate conscious visual experience: Evidence from inter-hemispheric summation

    Neuroreport

    (2001)
  • B. de Gelder et al.

    Non-conscious recognition of affect in the absence of striate cortex

    Neuroreport

    (1999)
  • Cited by (145)

    • Pretreatment Alterations and Acute Medication Treatment Effects on Brain Task–Related Functional Connectivity in Youth With Bipolar Disorder: A Neuroimaging Randomized Clinical Trial

      2022, Journal of the American Academy of Child and Adolescent Psychiatry
      Citation Excerpt :

      Compared with the traditional seed-based functional connectivity analysis usually used in resting-state fMRI studies, wSBC uses condition-specific weights defined from the hemodynamic response function. The wSBC was analyzed using atlas-based seed regions believed to be involved in BD pathology, including bilateral amygdala, ventral lateral prefrontal cortex (VLPFC), dorsal lateral prefrontal cortex, orbital frontal cortex (OFC), and insula,34 utilizing the WFU_PickAtlas MATLAB toolbox (https://www.nitrc.org/projects/wfu_pickatlas), leading to 10 separate seeds for connectivity analysis. The wSBC map was generated by computing condition-specific Pearson correlation coefficients between the blood oxygen level–dependent time series of the seed and the blood oxygen level–dependent time series of voxels outside the seed using a general linear model (GLM).

    View all citing articles on Scopus
    View full text