PER CURIAM.

We have on appeal the trial court's denial of a motion for postconviction relief pursuant to Florida Rule of Criminal Procedure 3.850. We have jurisdiction. Art. V, § 3(b)(1), (9), Fla. Const. The facts of the murder for which Terry Melvin Sims was sentenced to death are stated in the earlier direct appeal. Sims v. State, 444 So.2d 922 (Fla. 1983), cert. denied, 467 U.S. 1246, 104 S.Ct. 3525, 82 L.Ed.2d 832 (1984). In this proceeding, Sims raises a number of issues.

First, Sims argues that defense counsel was prejudicially ineffective during the guilt phase of the trial. This contention centers around the testimony of three eyewitnesses who were subjected to hypnosis on January 4, 1978, by a police officer, allegedly resulting in improved memory of events surrounding the killing. The hypnosis took place a few days after the murder when the police were endeavoring to get a better description of the killer. The hypnotist, Bruce Drazen, had taken several courses in investigative hypnosis but had no formal educational training in hypnosis at the time. He had been conducting hypnotic interviews for the police department for about three years. At the time of the hypnotic session, Drazen had not read the statements of the witnesses and did not know what the suspect looked like. During hypnosis, the main thrust of his questions was to try to develop some details to help identify the killers. Either during or after the hypnotic session, the witnesses talked to a police artist who attempted to draw pictures reflecting the descriptions of the killer's characteristics. One of the witnesses identified Sims in a photo lineup which took place about a month later, and all three of them identified Sims at the trial.

Before trial, defense counsel filed a motion to strike the introduction of the hypnotically refreshed testimony but apparently did not follow up on the motion. The witnesses were not cross-examined about having been hypnotized. At the postconviction hearing, one of Sims' lawyers stated that he and his cocounsel had not consulted with any experts concerning hypnosis and had done no legal research on the subject. He testified that there was no trial strategy behind his failure to present expert testimony on hypnosis and conceded prejudicial ineffectiveness. However, Sims' cocounsel said that he had conducted research on the subject and concluded that the testimony was admissible.

The hypnotic sessions had been recorded, but the tape of only one witness was available. However, it was conceded that the technique used by Mr. Drazen was similar with all witnesses. Drazen testified at the postconviction hearing that he tried to help his subjects relax so that they could remember more details. He testified that he used the Reiser Screen Technique developed by Dr. Martin Reiser for the Los Angeles Police Department "whereby you could visualize, you could picture things in the mind's eye, imagine you're looking at a movie screen, that the subconscious was . . . maybe the subconscious was a projector, and you would then project the information onto the screen. You could move the action forward or backward, freeze the action on the film projector and zoom in on what you're looking at." He agreed that in employing this technique there was a possibility of confabulation, which he defined as a mixture of fact and fantasy. Because of this, he said that any information obtained by hypnosis should be corroborated.

Dr. Buckhout, a psychology professor, testified for Sims at the postconviction hearing. He stated that he originally believed that memory was the result of the camera process in which people form pictorial reminiscences of what they have seen. He explained that the premise for forensic hypnosis assumes that you can get people to look at a screen and review the memory that they did not have before and suddenly reproduce it. However, he later became convinced that memory does not work this way. He now believes that the memory which is obtained through hypnosis is largely a product of suggestion. He admitted that in 1978, there were two schools of thought — that which believed in using forensic hypnosis to produce identification, and those who felt that hypnosis was really only useful in therapy. However, he said that in 1979 and 1980, experiments were run which demonstrated that hypnosis did not improve memory and that by 1987 there was a psychological consensus that hypnosis does not improve memory. He had listened to the available tape of the hypnotic session in this case and expressed the opinion that Drazen's hypnosis was unreliable. He agreed, however, that Drazen had not asked leading questions.

In rejecting Sims' claim on this issue, the trial judge stated in his order:

The record does not reflect why the State thought it to be necessary to use hypnosis in this case. Such a procedure was subsequently discredited and hypnotically refreshed testimony is no longer admissible. Bundy v. State, 471 So.2d 9 (Fla. 1985). However, this type of procedure was allowable at the time of the trial of this case and defense counsel determined that the credibility of the three witnesses who had been hypnotized could be successfully attacked by other means. This decision falls within the wide range of professionally competent assistance that counsel must make in every trial. Furthermore, counsel is not required to anticipate changes in the law resulting from subsequent court decisions. See Spaziano v. State, 489 So.2d 720 (Fla. 1986); Stevens v. State, 552 So.2d 1082 (Fla. 1989). The suggestion made by the defendant that the prosecutor placed the witness Guggenheim in a "post hypnotic trance" by asking him to recall the events of the homicide is rejected as being specious.

Finally, the defendant claims that reasonably effective counsel would have educated the jury on the dangerous unreliability of hypnotically refreshed testimony. This decision also falls within the wide range of decisions which counsel must make in representing a defendant. Dwelling on such an issue could result in an unwanted result. Jurors could have believed that hypnosis bolstered the credibility of an eye witness. Counsel should not be faulted for not taking that chance.

In gauging the ineffectiveness, of course, the United States Supreme Court in Strickland v. Washington, 466 U.S. 668, 104 S.Ct. 2052, 80 L.Ed.2d 674 (1984), required that a defendant show that trial counsel, first, "made errors so serious that counsel was not functioning as the `counsel' guaranteed the defendant by the Sixth Amendment." Id. at 687, 104 S.Ct. at 2064. Second, a defendant must also show "that counsel's errors were so serious as to deprive the defendant of a fair trial, a trial whose result is reliable." Id. In other words, the petitioner must prove actual prejudice, which consists of a "reasonable probability that, but for counsel's unprofessional errors, the result of the proceeding would have been different." Id. at 694, 104 S.Ct. at 2068. A "reasonable probability" consists, not of an absolute certainty, but of "a probability sufficient to undermine confidence in the outcome." Id.

The standard elaborated above is a strict one. As applied to this case, we believe there was sufficient evidence in the record to support the trial judge's ruling that counsel did not render ineffective assistance. In 1978, it was evident that forensic hypnosis was often employed. It was not until seven years later that this Court held that hypnotically obtained testimony was inadmissible. Bundy v. State, 471 So.2d 9 (Fla. 1985), cert. denied, 479 U.S. 894, 107 S.Ct. 295, 93 L.Ed.2d 269 (1986).

In any event, even if it could be said that counsel was ineffective, Sims was not prejudiced within the meaning of Strickland. In the first place, we find it significant that when Drazen was conducting his interview he did not know what the killer looked like. Therefore, his questions were not designed to lead the witnesses to a particular conclusion. Further, assuming that none of the hypnotized witnesses had testified, the record still would contain the testimony of two of Sims' accomplices, who had turned state's evidence. Their testimony included the strongest identifications in the record, since both stated that Sims admitted his guilt within seconds of the killing. Moreover, even if the Bundy analysis had been available and applicable to this case, much of the testimony of the hypnotized witnesses still would have been admissible; and this remaining testimony corroborated the statements of the two codefendants. Thus, while we recognize the defense efforts to impeach the testimony of the codefendants, we cannot agree that the record demonstrates a reasonable probability that the result of the proceedings would have been different.

Sims also argues that trial counsel failed to object when Sims was brought into court wearing shackles. However, Heffernan stated that it was possible the jury did not see the shackles, and he did not want to call everyone's attention to their existence. This was a reasonable strategic decision. Accordingly, we cannot say that counsel's failure to object constituted an error or omission under Strickland.

As his next issue, Sims argues that fundamental constitutional law was violated by the use of the hypnotically refreshed testimony in this instance. We find no merit. The exclusion of hypnotically refreshed testimony announced in Bundy was not premised on constitutional grounds, but on the possible unreliability of such evidence. Bundy, 471 So.2d at 18. The mere fact that some such unreliable testimony may have been admitted does not rise to the level of constitutional error where, as here, there is no reasonable probability that the verdict would have been different absent this testimony.

Sims also contends that prosecutorial misconduct tainted his conviction. Specifically, the prosecutor and one of the codefendants, B.B. Halsell, at several points either suggested or stated that the latter had accepted a plea bargain of ten years in prison. In actuality, the bargain was for a maximum of ten years' incarceration. Halsell's sentence — which was only two years — was not imposed until after Sims' trial was completed. Reading the record in its entirety, however, we do not find that there was any deliberate deception or prejudice. While some misstatements were made, we do not believe they constituted error or that counsel was ineffective in failing to object to the misstatements.

Next, Sims argues that the State deliberately withheld exculpatory evidence consisting of a document that allegedly would have connected another man, Terry Wayne Gayle, to the present crime. We find this argument highly improbable. Even accepting the defense's argument at face value, the document would have done no more than show that Gayle was copurchaser of "lock pullers" later used by Sims and his codefendants to steal a car. We see no way this document could have established either that Gayle was at the scene of the instant crime or that he actually committed the murder. The admissible evidence overwhelmingly established Sims as the triggerman; and no plausible evidence showed that Gayle was at the scene. This document thus was immaterial evidence, and its exclusion was not error. United States v. Bagley, 473 U.S. 667, 682, 105 S.Ct. 3375, 3383, 87 L.Ed.2d 481 (1985).

As his next point, Sims argues that the trial court's handling of the penalty phase violated Hitchcock v. Dugger, 481 U.S. 393, 107 S.Ct. 1821, 95 L.Ed.2d 347 (1987). In reviewing the 3.850 motion, the trial court below conceded error but found it harmless under the standard announced in Delap v. Dugger, 513 So.2d 659 (Fla. 1987). We agree. The instruction admittedly was not a model of clarity, but it did tell jurors they could consider the statutory mitigating factors "among others." The trial court's written order obviously also could have been clearer. We agree that there was other nonstatutory mitigating evidence that should have been weighed, but it is insubstantial compared to the aggravating factors. We specifically reject Sims' argument that his codefendant's lesser sentences constituted a mitigating factor, since the evidence shows that Sims was the triggerman. Having read the entire record, we find the Hitchcock error harmless beyond a reasonable doubt.

Next, Sims argues that trial counsel was ineffective during the penalty phase on a number of grounds. The first subissue deals with victim impact evidence. We agree with the trial court below that the statements of Sharon Mathis did not constitute victim impact evidence. We also agree that the prosecutor exceeded the bounds of proper argument, especially when he suggested that jurors consider what the victim might have said had he been able to testify. This in effect constituted an improper "Golden Rule" argument because it asked jurors to place themselves in the victim's place. However, it is equally clear that the improper remarks did not become a feature of the trial, and there is no reasonable probability that the outcome would have been different had counsel shown greater diligence in raising objections. See Strickland, 466 U.S. at 694, 104 S.Ct. at 2068.

We find no error in trial counsel's failure to ascertain the existence of other mitigating evidence so it could be introduced in the penalty phase. Sims had directed defense counsel not to collect this evidence. Counsel certainly has considerable discretion in preparing a trial strategy and choosing the means of reaching the client's objectives, but we do not believe counsel can be considered ineffective for honoring the client's wishes. There also was no error in failing to object to instructions informing the jury that its role is advisory in the penalty phase. Nor was there error in failing to tell the jury that a life recommendation means a minimum of twenty-five years in prison without possibility of parole. We agree it was error for the trial court to admit the unsubstantiated common law robbery conviction to aggravate the penalty. However, we do not find that counsel was ineffective, because there was no prejudice here. Sims had committed a separate, substantiated crime that would support aggravation — a 1971 assault with intent to commit robbery.

Finally, we reject Sims' claims that guiltphase errors mandate that the death penalty be vacated; and that fundamental error occurred when the trial court aggravated the penalty based on the common law robbery conviction. The former claim is without merit, and the latter was not fundamental error because Sims had committed a separate, documented violent crime sufficient to support the trial court's finding of aggravation.

For the foregoing reasons, the trial court's order denying relief pursuant to rule 3.850 is affirmed.

It is so ordered.

SHAW, C.J., and OVERTON, McDONALD, GRIMES and HARDING, JJ., concur.

KOGAN, J., dissents with an opinion, in which BARKETT, J., concurs.