The involvement of the frontal lobes in cognitive estimation

Download 59 Kb.
Size59 Kb.


Tim Shallice and Margaret E. Evans

(The National Hospital, Queen Square, London)

Clinical observations of patients with frontal lobe lesions have been used since the end of the nineteenth century in order to ascribe high-level cognitive functions to the frontal lobes, particularly its dorsolateral region (e.g. Jackson, 1884; Brickner. 1948; Ackerley. 1964; Luria, 1966). However there has also been an opposing view that has held that the frontal lobes had no primary focal role in complex cognitive operations. In the period after the Second World War, this latter view gained ground. being primarily based on findings that IQ test scores can be relatively unaffected after quite severe frontal damage (e.g. Hebb, 1945; Mettler, 1949; Teuber, 1964). Earlier clinical findings of intellectual deficits were attributed to the size of the lesions which were assumed to extend outside the frontal lobe (e.g. Teuber, 1955).

Clinical reports of frontal lobe cognitive deficits have, however, stressed a failure in judgement” and in dealing with novel situations while lower level cognitive skills are retained Grunthal, 1936: Goldstein, 1936). Indeed the most detailed recent theorising based on clinical observations --that of Luria (1966)-- may be summarised as attributing to the frontal lobes the functions of the selection and regulation of plans. A patient. J.S. recently assessed in the Psychology Department, National Hospital. gave further clinical support for a frontal locus for this type of deficit. As a result of an explosion he had sustained a massive right frontal lesion. But it appeared at operation that all the other lobes remained undamaged His IQ (verbal WAIS 89; performance WAIS 94; Raven’s Matrices 100) was on the same level as Army assessment arithmetic and vocabulary tests done before the injury was sustained (90 and 89 respectively). The only deficit observed was a gross inability to produce adequate cognitive estimates. For instance, when questioned, he replied that the height of the highest building in London was between 18,000 and 20,000 feet, that the largest fish in the world was a 3 foot long trout, that the best paid occupation was that of a long distance lorry driver, that the number of cars in Britain was over 50,000 and that the length of the average spine was between 4 and 5 feet. He did not appear to realise that the answers he give were bizarre and he justified them when pressed. When it was pointed out to him that the height he had given for the highest building was greater than the estimate he had given earlier of 17,000 feet for the highest mountain in Britain he merely reduced his estimate for the building to 15,000 feet.

Answering such questions satisfactorily stresses the abilities of select­ing an appropriate cognitive plan and of checking any putative answer obtained as much as the ability of carrying out the selected plan. Thus to obtain a reasonable estimate of the length of the spine the subject must utilise items of common knowledge in a relatively novel fashion but need not perform any complex computation. For instance, it may be obtained by working out the length of the body minus the head and the legs or by visualising the length of a jacket. Moreover if the calculation provides an unreasonable answer this can be checked against further everyday infor­mation. Thus a normal subject who by some means or another produced an initial estimate of the length of the spine as 4 1/2 feet would immediately see its absurdity, pointing out, for instance, that some people are little taller than that.

A group study was carried out to assess whether questions of this type do in general present especial difficulty for patients with frontal lobe lesions In addition the issue of whether performance on such questions can be dissociated from that obtained on more standard intelligence tests was investigated.


The test was given to all right-handed patients admitted to the National Hospital between January 1974 and September 1975, clinically assessed at the time as having a unilateral focal cortical lesion confined to no more than two lobes (throughout in assessing number of lobes involved, the occipital and parietal lobes were not differentiated), fit enough to leave their ward and clinically free to do so. Items 26-50 of set B of the Peabody Picture Vocabulary test was used as a screening test a written version). If the patient made more than 5 mistakes in naming the written words or matching them to the picture, he or she was excluded. (On the auditory version of the Peabody satisfactory performance at this level is to be expected with a mental age of 5.) Three patients were excluded, two for failing to read the words and one for failing to match appropriately.

Patients’ medical records were independently examined retrospectively to assess whether there had indeed been a unilateral localised lesion and if positive evidence of a lesion was not available from operation notes, at least three positive results were required using existing EMI Scan, Carotid-angiogram, Gamma Scan, Skull X-Ray or LEG evidence. Of the original 113 patients tested, 17 were rejected, seven because there was insufficient evidence of a lesion, two because the lesion was not localisable, four because the lesion was too extensive, one because it was sub-cortical and three because states of altered consciousness or epileptic attacks occurred during testing. Of the remaining 96 patients, 45 (average age 46.0) had left hemisphere lesions and 51 (average age 49.8) had right hemisphere lesions.

Patients were divided into four groups according to lateralisation and location of lesion. Any patient with frontal lobe involvement was assigned to the Anterior groups: patients without frontal lobe involvement were assigned to the Posterior groups. The majority of the patients had cerebral tumours. There was a significant difference in etiology between the Anterior and Posterior groups when meningiomas, intrinsic tumours and non-tumours were separated (chi = 16.93: p < 001; 2 d.f.). Visual field defects were significantly more common in the right than in the left hemisphere group (chi = 842: p < .01; 1 d.f.). Clinical details are summarised in Table I.

---Table I about here---
Twenty-five patients with extra-cerebral lesions (mean age 44.8) were tested as controls.


Twenty-six questions were devised such that an appropriate plan for answering them was not immediately apparent, but when obtained no specialist knowledge was necessary in order to achieve a reasonable answer. They were given to a preliminary group of control subjects. The fifteen questions which produced the smallest variance among the replies were selected from among them. The complete list of questions is given in the appendix.

Patients were given the following instructions: “Would you please make the best guess you can in answer to each of these questions. Almost certainly you will not know the exact answer: for instance, the height of a double-decker bus is l4 feet. but a reasonable guess would be anything from 12 to 18 feet.” Patients were given as long as they wanted to answer a question and if requested the question was repeated. Patients were encouraged to give an answer even if they initially said the:.’ had no idea.

Tests of arithmetic and non-verbal reasoning were also administered. The arithmetic test involved ten problems of graded difficulty to assess their ability to perform mechanical arithmetic operations (addition and subtraction). (In 6 cases performance on this test was estimated from the Arithmetic sub-test of the WAIS using an empirically obtained formula.) 79 of the patients with focal lesions also did sections B and C of Raven’s Matrices; it was not possible to test the other 17, for reasons such as the patient’s discharge.

A patient unable to obtain an appropriate strategy for answering a question or who has inadequate error-checking is more likely to produce a very incorrect response. It was therefore necessary to devise a measure of how bizarre a response was. The performance of the control patients was accordingly used to derive a four-point scale of the extremeness of the response (normal, quite extreme, extreme and very extreme). This scale was then applied to the results of the patients with cerebral lesions.

For most questions responses could be greater or smaller than the correct answer. Considering responses greater than the correct one, any answer greater than any of the estimates given by the control group was rated as very extreme, other answers greater than all-but-one of the control group’s ones were rated as extreme and those greater than all-but-two as quite extreme. Responses less than the correct answer were rated by a similar method. On certain questions (Qs 4, 8 and 12) where the question asks for the largest item in a given category, only deviations in one direction are theoretically possible. In these cases to maintain the same rate of bizarre responses among the controls, the very extreme, extreme and quite extreme ranges were ended at the smallest, the third smallest and fifth smallest responses (all inclusive) respectively. In the case of Q. 15 only answers of zero were treated as not normal; they were rated quite extreme. Any omission (only 0.5% of all responses made by the lesion groups) was treated as extreme. No responses were formally incorrect such as responding ‘tree’ as the largest object normally found in a house, except for the very frequent response ‘whale’ to Q. 8; this was rated normal as it was given by 64% of the controls.
---Table II about here ---
The results were analysed by means of a number of analyses of covariance (see Table II). As the use of any particular level of extremeness of response as a cut-off would be arbitrary, all analyses were repeated using as dependent variables the number of responses rated quite extreme or more so for each patient (measure I), the number rated extreme or very extreme (measure II) and the number rated very extreme (measure III). For every analysis the three measures gave similar results (see Table II); they will not therefore be separately discussed.

The first main result, shown in Table III, is of a frontal defect on the task. There is a significant effect of location (i.e. anterior, posterior) (see Anal sis U, but no effect of hemisphere and no interaction (F < 1.3 for all measures). Individual comparisons between the four lesion groups produce borderline significant differences between the Left Anterior and Left Posterior groups (see Analysis II), but no significant differences between the other groups (see Analyses III and IV).

The method adopted of allocating patients to groups means that the Anterior group contained both the fronto-temporal and fronto-parietal lesion sub-groups, and might therefore be expected to consist of patients with larger lesions. The analysis was therefore repeated excluding patients having damage to more than one lobe. The frontal deficit was, if anything, greater (see Analysis V). The different composition of the Anterior and Posterior groups in terms of aetiology can also be discounted as a
Percentage of Extreme Responses for each Cerebral Lesion Group

























possible source of artefact. Patients with meningiomas. which occurred more frequently in Anterior cases, have a tendency to produce a smaller not a greater score than patients with non-tumour aetiologies. which occurred more frequently in Posterior cases.

The second main result is that the frontal deficit cannot be attributed to a defect in reasoning or “general intelligence” as when this factor is partialled out by using Raven’s Matrices as a covariate, the frontal deficit on Cognitive Estimation is, if anything, more clearcut (Analysis VI). In addition, partialling out the effect of different levels of arithmetic skills and concepts of number by using the results of the arithmetic test as a covariate has a similar effect (Analysis VII).


The Anterior group performed significantly worse than the Posterior group on the Cognitive Estimation task. Moreover the deficit can be dissociated from one of “general intelligence” or reasoning as covarving the results with Raven’s Matrices scores left the effect unaltered.

Answering a Cognitive Estimate question such as “How long is an average man’s spine?” involves comprehending the question, selecting an appropriate plan for tackling the question, carrying out the plan using available cognitive skills and possibly checking that the answer is not too bizarre before producing the response. To attribute a focal difficulty on a task to any particular stage in the process, one must at least attempt to show that other stages are unlikely to be responsible.

As the response produced was always in the correct category (except for the very rare omissions), comprehension failures would seem most unlikely to be the main cause of the effect. Moreover if this were the case, left non-frontals would be expected to have much more difficulty than they actually do. Many of the problems require some limited arithmetic competence and indeed a measure of arithmetic ability does correlate with performance on the Cognitive Estimation task. Yet when this has been partialled out the frontal deficit remains. Various candidates remain as potential explanations of the deficit. In answering Cognitive Estimate questions more general knowledge is required than. say, for solving Raven’s Matrices; work on semantic memory, such as that of Warrington (1975) would, though, suggest a more posterior locus for the systems mediating it. A more specific deficit related to estimation or the non­-mechanical use of ordinally defined information could also conceivably be involve d.

The theoretical position by which the deficit is most simply explained is that of Luria (1966), that the selection and regulation of cognitive planning is one of the main functions of the human frontal lobes. Such planning functions would presumably by mediated through high-level programs which control the operation of lower-level cognitive programs themselves more posteriorly sited. Distinctions between different levels of cognitive programs have been made for both empirical and theoretical reasons, for instance by Broadbent (1977) and Sloman (1978).

On this view routine motor skills and routine cognitive skills such as the performing of mechanical arithmetic calculations would mainly require the use of only the lower-level programs. Even conventional intelligence tests. where a series of problems of the same type is presented with gradually increasing difficult, seem to demand the use of relatively routine even though complicated cognitive operations. This implies that such tests would stress primarily lower-level cognitive systems. Cognitive Estimate questions, by contrast, would stress the more anterior higher level system.
At present, however, this explanation of the obtained dissociation between performance on the two types of tasks remains speculative.

Ninety-six patients with localised cerebral lesions were tested on a task of providing reasonable answers to Cognitive Estimate questions. These questions are ones that can be answered using general knowledge available to almost all objects, but for which no immediately obvious strategy is available. It was found that patients with frontal lesions gave significantly more bizarre answers than patients with more posterior lesions. This effect is interpreted in terms of Luria’s (1966) theory of the planning functions of the frontal lobes.

Acknowledgements. We would like to thank the physicians and surgeons of the National Hospital for permission to study and report our findings on patients under their care.

We are grateful to Mr. J. Stevenson for his help in the localisation of the lesions, to Ms. E. Paul for her help with computer programming and to Dr. H.A. Buchtel particularly for his comments on the manuscript. Our especial thanks are due to Dr. Elizabeth Warrington for her assistance throughout the course of the investigation. The research was supported by a grant from the Medical Re­search Council.


The 15 questions with the percentage of bizarre responses (average of the three measures) for each lesion group were:

Cognitive Estimate Questions Left Right Left Right

ant, ant, post. post.

1. On average how many TV programmes are there on

any one TV channel between 6:00 pm and 11.00 pm? 16 21 21 16

2. What is the height of the Post Office Tower? 24 21 15 10

3. How fast do race horses gallop? * 37 25 21 25

4. What is the best paid job or occupation in Britain

today? ** 33 33 18 25

5. What is the age of the oldest person in Britain today? 25 29 13 32

6. What is the length of an average man’s spine? 22 17 14 12

7. How many slices in a sliced loaf? 8 10 7 7

8. What is the largest fish in the world? 11 3 3 11

9. How tall is the average English woman? *** 22 22 4 9

10. How heavy is a full pint bottle of milk? 33 21 26 15

11. How long is the average tie? 19 8 17 6
12. What is the largest object normally found in a

house? * 22 25 14 11

13. What is the width of a double.decker bus? 13 10 8 6

14. What is the length of a pound note? 19 7 29 14

15. How many camels are there in Holland? 27 19 17 19
Significance levels on measure 1: ***.01, **.05, * .1.


ACKERLEY, S. S. (1964) A case of paranatal bilateral frontal lobe defect observed for thirty years, in The Frontal Granular Cortex and Behavior, ed. by J. M. Warren and K. Akert, McGraw-Hill, New York.

BRICKNER. R. M. (1936) The Intellectual Functions of the Frontal Lobes: A Study Based Upon Observation of a Case of Partial Bilateral Frontal Lobectomy. Macmillan. New York.

BROADBENT, D. E. (1977) Levels, hierarchies and the locus of control. "Quart. J. Exp. Psychol.." 29, 181-201.

GOLDSTEIN. K. (1936) The significance of the frontal lobes for mental performance. "J. Neurol. Psychopathol.,” 17, 27-40. -

GRUNTHAL, E. (1936) Das Erkennen der traumatische Hirnverletzung. “Abh. Neurol..” Heft 76.

HEBB. D. 0. (1945) Man’s frontal lobes: a critical review. “Arch. Neurol. Psych.." 54. 10-24.

JACKSON. 1. H. (1884) Evolution and dissolution of the nervous system, in Selected Writings. vol. II. Hodder. London.

LURIA. A. R. (1966) Higher Cortical Functions in Man. Tavistock. London.

MCFIE, J., and THOMPSON. J. A. (1972) Picture arrangement: a measure of frontal lobe function? “Brit. .T. Psychiat.,” 121, 547-552.

METTLER. F. A. (1949) Selective partial ablation of the frontal cortex. Hoeber. New York. MILLER. G. A..GALANTER. E., and PRIBRANI. K. H. (1960) Plans and the Structure of Behavior.

Holt. Rinehart and Winston. New York.

SLOMAN. A. (1978) The Computer Revolution in Philosophy. Harvester. London.

TEUBER, H.-L. (1955) Physiological psychology, "Ann. Rev. Psychol.." 6,267-296.

— (1964) The riddle of frontal lobe function in man, in The Frontal Granular Cortex in Behavior. ed by. J. M. Warren and K. Akert. McGraw-Hill. New York.

WARRINGTON. E. K. (1975) The selective impairment of semantic memory, “Quart. T. Exp. Psvchol.,” 27, 635-657.

Tim Shallice. Psychology Department, National Hospital, Queen Square, London WC1N 3BG England.

Cortex (1978) 14, 294-303.

Share with your friends:

The database is protected by copyright © 2019
send message

    Main page