Grodzinsky's theory of agrammatic sentence processing fails to account for crucial empirical facts. In contrast to his predictions, the data show that there are (1) degrees of severity and (2) problems with sentences that do not require movement, and that (3) under the right task circumstances, full-fledged syntactic trees are constructed.

Localist networks represent information in a very simple and straightforward way. However, localist modelling of complex behaviours ultimately entails the use of intricate “hand-designed” connectionist structures. It is, in fact, mainly these two aspects of localist network models that I believe have turned many researchers off them (perhaps wrongly so).

Most current theoretical models of dreaming are built around an assumption that dream reports collected on awakening provide unbiased sampling of previous cognitive activity during sleep. However, such data are retrospective, requiring the recall of previous mental events from sleep on awakening. Thus, it is possible that dreaming occurs throughout sleep and differences in subsequent dream reports are owing to systematic differences in our ability to recall mentation on awakening. For this reason, it cannot be concluded with certainty that sleep cognition is more predominant or in any way different during REM compared to NREM sleep. It is our contention that REM sleep and ponto-geniculo-occipital (PGO) waves do not necessarily represent “pseudosensory” stimulation of the cortex in the generation of dreams, but might rather represent enhanced arousal of attention mechanisms during sleep, which results in the subsequent recall of attended mentation on awakening.

[Hobson et al.; Nielsen; Revonsuo; Solms; Vertes & Eastman]

Extremist views on normative rationality fail to address differences in responding owing to intellectual ability or epistemic self-regulation. Individual difference research thus raises serious questions concerning the scientific utility of universal rationality and universal irrationality theories. However, recent data indicate that computational capacity theories do not account adequately for within-subject variability in normative responding, memory-reasoning independence, and instances of ability-normative reasoning independence.

The explicit system for action selection integrates emotional information with the higher-order cognitive processes which culminate in the language system. Even the basic feels of emotion are what they are because they are integrated into the higher cognitive processes. The relation between emotion and intelligence would become increasingly important as the focus of brain science shifts to the integrative function of the prefrontal lobe.

Norris, McQueen & Cutler provide two possible explanations for neighborhood effects. The first suggests that nonwords that are more similar to words tend to activate those words more than do less similar nonwords, and the second is based on sequential probabilities between phonemes. Unfortunately, neither explanation is sufficient to explain all reported neighborhood effects.

Rolls defines emotion as innate reward and punishment. This could not explain our results showing that people learn faster in a negative mood. We argue that what people know about their world affects their emotional state. Negative emotion signals a failure to predict negative reward and hence prompts learning to resolve the ignorance. Thus what you don't know affects how you feel.

It is a major move from the claim that the core linguistic problem in Broca's aphasia is the inability to deal with traces, to the claim that this is the syntactic operation only and that it is exclusively supported by Broca's region. Three arguments plead against this move. First, many Broca patients have no damage to Broca's area. Second, it is not only passive, but also active jabberwocky sentences that activate the frontal operculum in a judgment task. Third, the same area is involved in a phrase-building production task that does not require tense processing.

The case for the independence of featural processing supports Merge and FLMP alike. The target article's criticisms of the latter model are founded on misunderstanding its application to natural language processing. In fact, the main difference in the functional properties of the models is the FLMP's ability to account for graded perceptual experience. @T3:It is startling to be informed that beliefs you have held for a quarter of a century are indefensible and implausible. So it comes as a relief to discover that it was all just a big misunderstanding.

Dream content is more coherent, consistent over time, and continuous with waking emotional concerns than most brainstem-driven theories of dreaming allow, but dreaming probably has no adaptive function. A new neurocognitive perspective focusing on the forebrain system of dream generation should begin with the findings on dream content in adults and the developmental nature of dreaming in children.

[Hobson et al.; Nielsen; Revonsuo; Solms; Vertes & Eastman]

Regardless of the clarity of the patterns they produce, individual differences in reasoning cannot validate norms of rationality. With improved reliability, these correlations will simply reveal which sorts of biases go together and which predict the intelligence of the decision maker. It seems necessary, therefore, to continue efforts to define rational thought independently of intelligence.

Page has done connectionist researchers a valuable service in this target article. He points out that connectionist models using localized representations often work as well or better than models using distributed representations. I point out that models using distributed representations are difficult to understand and often lack parsimony and plausibility. In conclusion, I give an example – the case of the missing fundamental in music – that can easily be explained by a model using localist representations but can be explained only with great difficulty and implausibility by a model using distributed representations.

Grodzinsky's Tree-Pruning Hypothesis can be extended to explain agrammatic comprehension disorders. Although agrammatism is evidence for syntactic modularity, there is no evidence for its anatomical modularity or for its localization in the frontal lobe. Agrammatism results from diffuse left hemisphere damage – allowing the emergence of the limited right hemisphere linguistic competence – rather than from damage to an anatomic module in the left hemisphere.

New findings point to a role for mesolimbic DA circuits in the generation of dreaming. We disagree with Solms about these structures having an exclusive role in generating dreams. We review data suggesting that dreaming can be interrupted at different levels of processing and that anterior-subcortical lesions associated with dream cessation are unlikely to produce selective hypodopaminergic dynamic impairments.

[Hobson et al.; Nielsen; Solms]

In their use of correlations as a means to distinguish between different views on the normative/descriptive gap, Stanovich & West discuss the competence component but neglect the activation-utilization component of performance. Different degrees of motivation may introduce systematic variation that is confounded with the variation explained by cognitive capacity.

Norris et al. present a sufficiency case for Merge, but not for autonomy. The simulations make clear that there is little reason to favor Merge over TRACE. The slanted presentation of the empirical evidence gives the illusion that the autonomous position is stronger than it really is.

Bifurcation analysis of a real-time implementation of an ART network, which is functionally similar to the generalized localist model discussed in Page's manifesto shows that it yields a phase transition from local to distributed representation owing to continuous variation of the range of inhibitory connections. Hence there appears to be a qualitative dichotomy between local and distributed representations at the level of connectionistic networks conceived of as instances of nonlinear dynamical systems.

Rolls shares important data on hunger, thirst, sexuality, and learned behaviors, but is it pertinent to understanding the fundamental nature of emotionality? Important as such work is for understanding the motivated behaviors of animals, Rolls builds a constructivist theory of emotions and primary-process affective consciousness without considering past evidence on specific types of emotional tendencies and their diverse neural substrates.

Grodzinsky presents interesting data on Broca's aphasia, but because of obsolete ideas about neurofunctional organization and an inadequate review of the neuroimaging literature, he fails to put these data into perspective. Rather than supporting a specific linguistic function of Broca's area, the findings should be viewed in terms of working memory functions of the inferior frontal cortex.

We urge caution in generalising from content words to function words, in which lexical-to-phonemic feedback might be more likely. Speech perception involves more than word recognition; feedback might be outside the narrow logic of word identification but still be present for other purposes. Finally, we raise the issue of evidence from imaging studies of auditory hallucination.